Automatic Coulometric Titrations Involving Amperometric End Point

again records the necessary' data, from which he can calculate the slope of the indicator current (against generation ..... A suitable filter (C¡g, R...
0 downloads 8 Views 678KB Size
Automatic Coulometric Titrations Involving an Amperometric End Point H E N R Y L. RICHTER', JR. California lnrtitote of Technology, Pasadena 4, Calif.

An instrument has been developed for automatic control of coulometric titrations employing the dual indicator electrode amperometric end point. The instrument is capable of detecting the end of the titration and stopping the generation, freeing the analyst of this task. The titration is stopped at a preset indicator current, either as a dead-stop end point or as a preliminary operation to determining the exact end point by extrapolation of the post-end-point current to zero. The operation of the instrument on three different types of coulometric titrations is described.

ilRIOUS coulometric titrations and the end points used in this laboratory have been described ( 2 , 4, 7-11), This paper is concerned with titrations that make use of dual indicator electrode amperometric. end points which have been used in this laboratory (see Figures 1 and 2). The dual indicator electrode amperometric end-point system forms a basis for a versatile tit,rator. The simplicity 'of the amperometric syst'em, both functionally as to the physical arrangement and in its direct application t,o a number of different titrations, is essential in a system where different types of analyses are to be performed. The fact that the response of the indicator circuit is almost instantaneous simplifies circuit design. The dual indicator electrode system has the addit,ional advantage that in most instances the indicator current does not change the concentration of the system being measured. The three types of amperometric end points which have been encountered and toward which the inst,rument has been designed are s h o w in Figure 1 ( A , B, and C). The most common type of end point is that represent'ed by curve A , where t.he indicator current has a steady or very slowly rising value unt,il the vicinity of the end point. iifter the end point is passed it increases rapidly in a linear manner (when plotted against generation time as shown in Figure 1). A representative end point, of this type occurs in the titration of arsenic with electrolytically generated bromine (6). Several authors describe automatic coulometric t,it,ratorswhich use this type of end point ( 1 , 5 , 6). An automatic coulometric titrator that employs potentiomet,ric detection of the end bf the titration has been described (5). Curve B represents the case where the end point is preceded by a sudden reversal of indicator current, as has occasionally been experienced in titrations involving electrolytically generated bromine and chlorine (6). This reversal is believed to be due to a shift a t an electrode from one controlling half cell to another. Curve C represents the indicator current in a third type of titration represent,ed by the titration of iodide with electrolyticallv generated bromine (9). I n t,his type of tit.ration, the end point is preceded by a high maximum of indicator current. The usual met,hod of locating the end point has been described in detail by several authors ( 2 , 6, 9 ) ) but essentially consists in all cases of extrapolating the linearly rising post-end-point current to zero current and then applying a correction obtained from a lilank titration. Thus, when used n-ith this amperometric technique, this instrument is. not required to stop the tit,ration esactly a t the end point, for the exact end point of a titration is obtained mathematically, as indicated. The instrument which hx.3 been developed nil1 stop a titration a t any preset level of 1 Present address, Jet Propulsion Laboratory, 4800 Oak Grove Drive, Pasadena, Calif.

indicator current within its range. The analyst then records the value of this indicator current and the elapsed generation time from the timer, generates manually for 1 second more, and again records the necessary data, from which he can calculate the slope of the indicator current (against generation time) for the purpose of extrapolating to zero end point current. He then applies correctiona obtained from a blank determination to find the corrected end point. Although a dead-stop type of instrument is not necessary for this technique, it would be possible to use the instrument here for routine analyses as a dead-stop instrument with accuracy on the order of 1%. If a number of similar titrations were being made, it would be feasible to calculate a factor (on the basis of a blank titration) to enable the analyst quickly to compute the results of the determination on the basis of the elapsed time as indicated directly by the timer.

too

LOO

QENERATION

Figuie 1.

TIML.

SEC.

Indicator current during three types of coulometric titrations

One other means of locating the end point would be to record the indicator current and use the automatic controller to stop the titration after the end point has been passed. The end point could be obtained from the recorded indicator current in the manner described above. PRINCIPLE O F OPERATION

The instrument described was constructed as an accessory unit which can easily be attached to existing coulometric titrators. The basic titrator ( 5 ) consists of a titration cell, an indicating circuit which employs two platinum electrodes in the titration cell, and the generation circuit. The latter consists of an electronically controlled constant-current power supply with an elapsed time indicator, in connection with two other platinum electrodes in the titration cell. For automatic operation, the progress of the titration is followed by measuring the current between the indicator electrodes and the potential drop across the electrodes (the latter having an inverse relationship to the magnitude of the indicator current). The indicator current is actually measured in the form of a potential created by the flow of the indicator current through a series resistor, R (Figure 2 ) . Because of the magnitude of the indicator current used in the dual indicator electrode amperometric method of indication described here (0 to 50 pa.) and the amounts of resistance that can be placed in series with the indicator electrodes (generally 1000 1526

1527

V O L U M E 27, NO. 10, O C T O B E R 1 9 5 5

the indicator current becomes more positive than the preeet negative value for V,, and V4 then terminates the titration as for end point A . End point C of Figire 1 requires the use of the potential drop across the indicator electrodes. The appropriate block diagram is Figure 3,c. Consider the circuit of Figure 2, which shows a resistance in series 1%-iththe indicator electrodes. The combined resistance of the sensing resistor, R, and the resistance of the indicator current meter is of such a value that it is small compared t o the resistance of the cell when very little indicator current is flowing and large compared to the cell resistance when 10 or more niicroamperes of indicator current are flowing. The

Figure 2.

Complete coulometric titrator

1-

C"0PP.I

Indicotci C u r r e n t

(a)

ohms), the potential available for control purposes is on the order of 0 to 50 mv. The instrument therefore must contain some sort of amplifier t o produce power sufficient to actuate control relays. The amplifier must be of the direct-current type, as the indicator current is of this classification, and the direction of flow of this current is of importance. The amplifier selected for this application is a so-called "chopper amplifier" which converts a highgain, stahle alternating current amplifier into a direct current amplifier. Block diagrams of the apparatus are shown in Figure 3.

v

8, and Vz constitute the direct current amplifier. V 3 and V4 are electronic relay circuits. V 4 is a cathode-coupled amplifiqr intended to give inversion of polarity, which operates on a positive input (normal flow of indicator current). Vacuum tubes T i 3 and V,actuate relays when the output of the amplifier reaches a certain preset value-that is, v-hen the indicator current or potential drop between the indicator electrodes reaches a certain value. The operation of the instrument on the various end points is discussed with the aid of the block diagrams of Figure 3. Titration curve A (Figure l), the simplest of the three, is described first in a simple way, and then after titrations B and C are discussed, titration curve A is described fully. Thus, first, circuit Va will be ignored. I n all three types of end points shown, the termination of the titration occurs on the post-end-point linear rise of indicator current. End point A consists only of this linear current rise. Relay tube V4 can be preset to stop the titration a t any desired value of indicator current-let us say 10 Ha. (Figure 3,a). The potential developed across the sensing resistor, R, is amplified by VI and V2and applied to V4. When it reaches the preset value, the relay in the plate circuit of V4 closes and the titration is terminated, I n actual operation with a titration made on a fast generation rate there may be a lag in the operation of the device and the final indicator current may be 15 pa. when shutoff was set for 10. Titration B is similar to A , but contains a sudden reversal of indicator current. When and if this occurs, it is necessary to stop the titration and give the titration system a few seconds to attain equilibrium conditions again. This reversal does not always occur. The appropriate block diagram is Figure 3,b. Relay tubes V 3and V4are both connected to the amplifier output. 8,again responds t o a positive flow of indicator current as was described for end point A. V 3responds to a negative flow of indicator current and is set to a lon- value. If a value of indicator current more negative than this is encountered, this circuit stops the titration. This can be of the form of a permanent termination of the titration (which is customary, as this occurs near the end point) or the machine can be set to resume the titration when

Y

(C) Figure 3.

Block diagrams of operation of automatic controller

potential drop across the indicator electrodes will not be constant, but, because of the external resistance, will be strongly dependent on the indicator current. This potential drop is shown qualitatively by the dashed curve, D, of Figure 1. When very little indicator current is flowing, this potential is almost the applied indicator electrode potential. When indicator current flows during the titration, because of the potential drop across the external resistance in the indicator circuit, the potential across the indicator electrodes decreases. If the same system were used for end point C as for A , relay tube Ti, would respond at the first rise of indicator current and the titrations would be stopped prematurely. To avoid this, V I is used in such a way as to prevent application of the amplifier output to V , until just before the end point. The relay associated with V 3is uded to switch the amplifier input from across the indicator electrodes to across the sensing resistor, R. A titration involving end point C begins with the amplifier input connected to the indicator electrodes, the potential

1528

ANALYTICAL CHEMISTRY

I

'

8

:

I

,

I

,

N n

K

n

Y

m

a

7 J m V

I +

3-

I

-+-I?a

I

.....

-I

Y

V O L U M E 27, NO. 10, O C T O B E R 1 9 5 5

1529

of which is chosen to he opposite in polarity to the potential across E . The amplifier output is always connected to both Va and VI and, because the output of the amplifier will be negative during the first part of the titration, VI will not respond to the signal. V3 does respond to a negative output and is preset to respond a t a potential near to the applied indicator potential. The desired point of response is just before the maximum on curve D of Figure 1, ahich occurs just before the end point. When this happens, the relay associated with V3 changes the input to the potential drop across R and the titration completes as was described for titration A . By the procesa just described, the sampling of the high intermediate indicator current by the amplifier is avoided. As Vr would respond to the potential across the indicator electrodes a t the start of the titration when it is high [Figure 1, D), a gate tube is used for V3, and is disabled during the first part of the titration, effectively producing a condition replacing the initial dashed portion of curve D with that part indicated by dots. TitrationsinvolvingendpointA can now beaccurately described, as the same arrangement is used as for end point C. The indicator electrode potential (when plotted against generation time) bears an inverse relationship to the indicator current (compare Figure 1, C and D). During a titration which culminates in an A type of end point, the indicator electrode potential differences will be high until the linear rise of current denoting the end point. Thus, after the disabling of 17, a t the ;Itart of the titration, V 3 nil1 respond a t once, owing to the high negative input, transferring the input to the amplifier to the potential drop aero93 R. The titration will then terminate in the manner described for end point A a t the beginning of this discussion. No changes in switching need to be made to differentiate between end points A and C. End point B requires a different arrangement, the poasibility of a current reversal is known beforehand and necessary awitching [from c, Figure 3, to b ) can he done. I n the titration? where indicator current reversals are sometimes experienced, intermediate indicator current maxima have not been found.

m

5

t h2

N

b

a d

h

3 c

m

*

F

Fi

u

B

$

ELECTRONIC CIRCUITS

m

w c

E

-T

5

i

11

With the instrument set for end point B , the input to the amplifier is always the potential drop across the sensing resistor, R, and the response of either Vs or V4 can terminate the titration. With the instrument set for titration C (or A ) the input to the amplifier is switched by the relay associated with V3 and only V , can terminate the titration. Provision has been made for more than one value of sensing resistor to be available. The sensing resistor may be either 500 or 10,000 ohms, depending on the conditions of the titration. If the indicator current and its pre-end-point maximum are small, RZ9(10,000 ohms) is used. The sensing resistance is selected by a switch, St. The schematic circuit for the instrument is presented as Figure 4 With the titrations that have been discussed above, two types of end-point control are available. I n the case of a titration involving a J o w rate of reaction, it may he desirable to have the instrument stop the generation when the preset value of indicator current is reached, but to resume the titration if the indicator current falls beloK the preset value of indicator current when the excess intermediate has reacted with the substance being titrated. On the other hand, it may be desirable to have the instrument permanently stop the titration when the preset value of indicator current has been reached. This control is accomplished by a switch, which connects holding contacts on the generation control relay, Vd, representing the "permanent stop" condition and when open is the "slow rate" position. VI and Vz of Figure 4 and associated components comprise the direct current amplifier. V1 and Vp are connected in the form of a high-gain audio-frequency amplifier and this is converted to a direct-current amplifier by means of the chopper indicated. The operation of this unit is not discussed here The direct

ANALYTICAL CHEMISTRY

1530

EXPERIMENTAL

Table I. Run 1

7 8

A 60 60 51 51 51 60 60 60

9 10 11

62

2 3 4 5 6

548

io, Ira. 1 0 1 1 1 1 1 1

E1

Titrations of Iodide by Bromine fend, Sec. it, Fa. i t + 1, Ira. T,Sec. 87.687.0” 85.52 85.0 85.6 86.0 85.92 86.2 85.8:

29 31 22 22 22 24.5 28 26 29.5 16 28

of iodide taken for each run

34 37 27 30 29.5 30.5 35 32 36.5 21 35

81.85 81.86

O;:;: 82.6; 81.88

;;.

0; 81.9; 81.58 81.88 .4v. found

S.d.

cwrent amplifier was found to be the most critical part of the assembly as far as construction was concerned, careful isolation and shielding are necessary to avoid pickup Instabilities were encountered and traced to two sources. The adjustment of the chopper is important, it is a make-before-break switch and GOUpling between the input and output of the alternating current amplifier occurs if the adjustment is otherwise. I t was also found necessary to bypass the plate circuits of V1 and V z with small condensers to avoid oscillation (Ci and CIS). V3 is used as a pox-er amplifier. The sensitive rela)- in the plate circuit actuates a larger relay which does the necessary switching Desensitization of V3 at the start of the titration is accomplished by the network in the suppressor grid circuit. When switch S4 is thrown from “manual” to “automatic,” condenser Cl4 (v,-hicli was charged to the supply voltage) discharges through R,,, vith a time constant such that is gated for about 10 seconds. During this time the plate current is zero and response of this stage to the output of the amplifier is avoided. The amplifier output is applied to the tube by means of R21, which in effect determines the indicator electrode potential, or negative indicator current a t which relay K , will close. V4 is a cathode-coupled current amplifier. The value of the cathode resistor is somewhat critical and may have to be determined by experiment to put the operation of the tube in the proper region of plate current. The maximum output of the chopper amplifier is several volts and this is fed to the grid of V4. The cutoff level is set by potentiometer RZ5in the grid of the second half of the tube, which applies a negative bias. This bias is obtained from a resistor, Rz,, in series with the negative supply connection in order to get a fairly constant voltage. A suitable filter (C,, R26, Cis) is provided to remove any ripple components that might be present. Ordinary constructional practice for wiring has been followed. Pains as to proper shielding on the high-gain and high-signal parts of the chopper amplifier must be taken. A shield acrow the chopper socket, to isolate the input and output leads of the highgain alternating current amplifier, is recommended. Nost component values are not critical. Although the sensitive relays in the plate circuits of J‘a and Vd were placed under the chassis in this instrument, an arrangement which would permit them to be accevsible for adjustment in the normal operating position would be very desirable, as their operation is greatly influenced by gravity. The amplifier used has more than ample gain, and some sort of feedback could be profitably included for the purpose of stabilization and prevention of amplifier saturation. It has not been attempted here. I n the interest of reducing transients which may be produced by switching, all relay contacts carrying 110volt current have been appropriately bypassed bv capacitors, which are not shown in Figure 4. A lever-type telephone switch, S2, changes the circuitry from end point B to A and C (Figure l), switching the input of the amplifier and the sequencing of Va and 8,.A push-button snitch, &, is used as a ‘‘clearing” switch to open all relays at the start of a titration

Several different types of titrations have been w,Y successfully performed with the automatic instru546 ment. The prime funct,ion of the instrument is 546 to stop the titration in the vicinity of the end 54 1 549 point. Blthough experimental results are pre551 547 sented, the accuracy of coulometric titrations in547 volving dual indicator electrode amperometric end 547 548 points determined as described above is not de544 547 pendent upon whether the titration is performed 546 by manual or automatic control of the generation. 2.5 The crucial test was that the titration was automatically stopped a t or near the proper postend-point current. Titrations of tripositive antimony by means of electrolytically generated bromine of the type described by Brown and Swift ( 2 ) have been carried out with the instrument described above by E. A. Butler of this laboratory. I n every case the titration was stopped a t the preset indicator current level, and the results of the titrations agreed within experimental error (0.2%) with the calculated amount of tripositive antimony taken. Titrations of thiosulfate with electrolytically generated iodine (9) were made by J. K. Rowley of this laboratory with both manual and automatic control of the generation. The instrument worked satisfactorily in this application, both methods of control giving similar results. To test this instrument on an end point such a3 represented by C of Figure 1, iodide was titrated by elPctrolytically generated bromine using both manual and automatic control. The titrations were done so as to duplicate the method of Wooster, Farrington, and Swift ( I I ) , except that by necessity here the indicator current was allowed to flow throughout the titration. The results of one series of titrations are summarized in Table I. Here A is the potentiometer setting xhich adjusts the shutoff level on V4, io is the indicator current a t the start of the titration, tend is the reading on the timer when the titration stops, il is the indicator current a t the time of stopping, and it + 1 is the indicator current after 1 second of manual generation after stopping. The calculated time of the end point is T,and JV is the calculated weight of iodide found. The correction from a blank determination was negligible. The accuracy and reproducibility of the results are comparable to those of Wooster, Farrington, and Swift (If) and independent of the setting of the potentiometer A . Before making titrations with the automatic controller, as with any piece of electronic apparatus, it is advisable to allow about a 20-minute warm-up period. It has been found necessary here to perform a couple of initial titrations a t the beginning of each new series of runs in order to establish the two potentiometer settings required for correct shut-off points. The use of the instrument for dead-stop titrations has not been explored. For a series of titrations carried out under similar conditions, it should be possible to calibrate the machine for dead-stop usage when maximum accuracy of the method is not required. ACKNOWLEDGMENT

The advice and interest of Ernest H. Swift, under whom this problem was started as undergraduate research, have been appreciated. The author would like to thank Bart Locanthi, now of Computer Engineering Associates, for advice on chopper amplifiers. LITERATURE CITED

(1) Bett, N,, Nock, W., and Morris, G., Analyst, 79, 607 (1954). (2) Brown, R. A , , and Swift, E. H., J. Am. Chem. Soc., 71, 2717

(1949). (3) Carson, W.N., Jr., A s . 4 ~ CHEM., . 25,226 (1953). (4) Farrington, P.S., Meier, D. J., and Swift, E. H., Ibid., 25, 591 (1953).

V O L U M E 2 7 , NO. 10, O C T O B E R 1 9 5 5

1531

Leisey, F. A , , Ibid., 26, 1607 (1951). Lingane, J. J., Ibid., 26, 622 (1954). Meier, D. J., Myers, R. J., and Swift, E. H., J. Am. Chem. Soc.,

(11) Wooster, W. S.,Farrington, P. S., and Swift, E. H., Ibid., 21, 1457 (1949).

71, 2340 (1949). (8) llyers, R. J., and Swift. E. H., Ibid., 70, 1047 (1948). (9) Kowley, Keith, and Swift, E. H., ANAL.CHEM.,26, 373 (1954). (10) Sease, J. W., Xemann, C., and Swift, E. H., Ibid., 19, 197 (1947).

RECEIVED for review April 1, 1955. Accepted June 29, 1955. Presented in part before the Division of Analytical Chemistry a t the 123rd Meeting of the AMEBICAX CHEMICAL SOCIETY, Los tingeles, Calif., March 1953. Contribution 1981, Gates and Crellin Laboratories of Chemistry, California Institute of Technology, Pasadena, Calif.

Precise Assay of Trichloroacetic Acid by Coulometry at Controlled Potential T H E L M A MEITES

and

LOUIS MEITES’

Sterling Chemistry Laboratory, Yale University, N e w Haven, Conn.

The reduction of trichloroacetate ion from an ammoniacal medium proceeds quantitatively to dichloroacetate ion at a mercury cathode whose potential is maintained at a constant suitable value. Integrating the current which flows during such an electrolysis permits the determination of trichloroacetate in pure samples or in the presence of at least 8 times as much dichloroacetate, with an accuracy and precision of within &0.29”.

M

and dichloroacetic acids are universal contaminants Oxoof trichloroacet,ic acid, and no method of assaying trichloroacetic acid ( 4 , .9) has yet been proposed, which is free from interference by the Ion-er chlorinated acids. Such a method is presented i n this paper. I t is hosed on the quantitative electroreduction of trichloroacet,ate ion to dichloroacetate ion in an ammoniacal medium, which proceeds with 100% current efficiency a t a mercury cathode a t a suitably chosen potential. An integration of the current flowing during this reduction gives the amount of trichloroacetate present in the sample to =!=0.2% or better. The reduction of the chloroacetic acids a t a mercury cathode has been studied several times by polarographic techniques. Elving and Tang ( 2 . 3)showed that dichloroacetate ion in ammonical media gives a single wave at a fnirlv negative potential, corresponding to the reaction C1,CHCOOH20 2e +. ClCH2C00C1OH-

+

+

+

+

Trichloroacetate ion under the same conditions gives a double wave: The second wave is identical with the single dichloroacet,ate wave, while the first wave represent’s bhe reduction of trichloroacetate to dichloroacetate. This first wave, which is well defined, was used by Elving and Tang ( 2 ) for the polarographic determination of trichloroacetate in the presence of dichloroacetate; of course this is insufficiently accurate for assay purposes. Elving and Tang (3) found that monochloroacetate ion was not reducible from an ammoniacal solution. Neiman, Ryabov, and Sheyanova (8),on the other hand, asserted that monochloroacetate does give a single wave in 0.lM sodium hydroxide, and that dichloroacetate and trichloroacetate correspondingly give two m d t,hree waves, respectively. This the present authors were unable t o confirm; the polarographic characteristics of the chloroacetic acids in sodium hydroxide media do not differ in any significant respect from those found in ammoniacal media by Elving and Tang (3). From this information it appeared possible to select conditions under which trichloroacetate ion alone could be reduced, and quantit,atively, and to apply this to the coulometric determination of trichloroacetate, either alone or in mixtures with dichloroacetate.

* Present address, Department of Chemistry, Polytechnic Institute of Brooklyn, 99 Livingston S t . , Brooklyn 1, X. Y .

EXPERIMENTAL

The double diaphragm cell, the potentiostat, and the current integrator, which are manufactured by Analytical Instruments, Inc., Bristol, Conn., have been described ( 6 ) ,as has the recording polarograph used ( 7 ) . All weights and volumetric apparatus were carefully calibrated by conventional techniques. The chloroacetic acids were secured from a commercial supplier. The trichloroacetic acid was dried for a week over anhydrous magnesium perchlorate. Titration of a sample of the dried acid with sodium hydroxide standardized against potassium hydrogen phthalate gave (on the assumption that no lower chlorinated acid was present) an “assay” of 99.98%. Stock solutions of the acid appeared to be stable indefinitely when animonium and potasJium chlorides were present alone, but when ammonia was also present the values secured by coulometric analysis decreased a t the rate of about 1% per week. The kinetics of this reaction have been st,udied by Verhoek ( I O ) . Some of the mixtures of di- and trichloroacetic acids were prepared from the commercial dichloroacetic acid. Analysis of this mat,erial, either polarographically ( 2 ) or by the coulometric procedure described below, indicated the presence of roughly 0.3% trichloroacetic acid. The necessity of correcting for this amount of impurity would have severely limited the accuracy attainable with the mixtures containing much dichloroacetic acid. Consequently, a pure dichloroacetate solution was prepared by the following procedure. A solution of 4.6 grams of trichloroacetic acid in the ammoniacal supporting electrolyte used throughout the work mas electrolyzed a t a mercury cat’hode a t -0.8 volt vs. S.C.E. until the current had fallen to zero, and the resulting solution was transferred to a 250-ml. volumetric flask and diluted to the mark with the stock supporting electrolyte. Whereas ammoniacal solutions of the commercial dichloroacetic acid rapidly yellowed on standing, the electrolytically prepared solution appeared stable for many mont,hs. RECOMMEh-DED PROCEDURE

Prepare a stock solution containing 2.5M ammonia, 131 ammonium chloride, and 211.1 potassium chloride. (The potassium chloride serves primarily to decrease the cell resistance; and none of the concentrations is in any way critical.) Dissolve a sample of trichloroacetic acid weighing 0.03 to 5 grams in a little water and add about 40 ml. of the supporting electrolj-te solution. If the weight of the sample exceeds 1 gram, enough additional ammonia should be added to restore that lost by neutralization. Fill the auxiliary electrode and central compartments of a double diaphragm cell for controlled potential electrolysis in the manner described previously ( 6 ) ,and transfer the trichloroacetate solution to the working electrode compartment. Since the time required to complete the electrolysis is proportional to the volume of the solution ( b ) ,not more than an additional 50 ml. o f the supporting electrolyte Yhould be used in completing the transfer. Pass a stream of tank nitrogen or hrdrogen through an efficient gas washing bottle filled n-ith the supporting electrolyte and containing 1 to 2 grams of hydrazine dihydrochloride to facilitate the removal of oxygen, and thence into the solution in the n.orking electrode compartment of the cell. Since oxygen is reducible a t a mercury cathode under the conditions used in this procedure, it is essential t o remove all dissolved air before the electrolysis is begun. This normally requires about 5 minutes, provided that a sufficiently rapid stream of gas is iised and the solution is well stirred. Bdd about 30 ml. of pure mercury, read the coulometer register,