NEWBERY: ELECTROMOTIVE FORCES IN ALCOHOL. PART V. 2553CCXXXVI1.- Electromotive Forces in Alcohol. Part V.The Dropping Electrode in Alcoholic Solutions.By EDGAR NEWBERY.MEASUREMENTS of the electromotive forces of concentration cellspromise to furnish the most satisfactory method of comparing thethermodynamic potentials of an electrolyte in two dilute solutionsin the same solvent, and in earlier papers of %his series such applica-tions with calomel and hydrogen electrodes in alcoholic solutionshave been dealt with.A comparison of the thermodynamic potential of an electrolyt2554 NEWBERY : ELECTROMOTIVE FORCES IN ALCOHOL. PART V.in one solvent with that of the same electrolyte in another solventcannot. a t present be made, previous attempts notwithstanding, byany known electrometric method.The use of concentration cellscontaining two diflerent. soivents must at present be whollyexcluded, as the potential difference a t the boundary cannot bedetermined directly, and cannot even be roughly estimated, as thecalculations involved must be based on theoretical assumptionswhich would have a purely speculative character.For this reason the work described in the present series of papers(Hardman, Lapworth, and Partington, T., 1911, 99, 1417, 2242;1912, 101, 2249 ; this vol., p. 2302) is being developed along severaldifferent lines, which it is believed may ultimately converge a t amore promising point of attack. It, is sufficient a t present toindicate that this point is considered to be the determination of thedifference in absolute potential of two electrodes, one in any suitrable aqueous solution, and the other in any suitable alcoholicsolution, and for this purpose it is clear t'hat the precise valuesof the absolute potentials are not required t o be known; moreover,the matter is greatly simplified if conditions can be devised sothat the absolute potentials of the two electrodes in question areidentical.Direct determination of the absolute potentials of electrodes insolutions in both solvents offers one possible line of investigation,and of methods hitherto suggested f o r such determinations thedropping electrode and the capillary electrometer are the onlydevices which it is necessary to consider.It is well known thatthere are great difficulties in reconciling the results obtained bythe use of these two instruments even in aqueous solutions, and$there are already reasons f o r supposing that the most rapid progressis likely t o be inadel by further study of the capillary electrometer.'The present communication furnishes additional evidence of theuntrustworthy character from the present point of view of measure-ments made with the dropping electrode, even when the resultsexhibit most perfect consistency, and in later communications i twill be shown that the capillary electrometer promises to providethe material necessary for bridging this gap in the applications ofelectrometric methods.The dropping electrode has been applied t o the determination ofabsolute potentials in aqueous solutions by Paschen (Ann.Phys.Chern., 1890, [iii], 41, 42) and by Palmaer (Zeitsch. physikal.Chern., 1899, 28, 259; Zeitsch. Elektrochem., 1903, 9, 754), andothers.The apparatus used in the experiments described in the presentcommunication was similar t o that used by Palmaer (ZeitschNEWBERY : ELECTROMOTIVE FORCES IN BLCOHOL. PART V. 2555E'lektrochem., 1903, 9, 755), but with the following modifica-tions :(1) Pressure on the mercury was obtained by connecting a largepear-shaped vessel with the mercury dropper by means of 450 cm.of canvas-lined pressure tubing, and raising this vessel when fullof mercury by means of a cord passing over a pulley.(2) The electrode vessel was arranged so that it could beimmersed in a thermostat a t 25O.(3) A small burette, B, holding about 2 c.c., was fitted t o theAtop of the electrode vessel, to allow for the addition of smallmeasured quantities of any required solut,ion without, disturbingthe apparatus.A slow current of pure hydrogen, obtained by the electrolysisof pure sodium hydroxide solution, was passed through the appara-tus for some hours before t'aking measurements, and continuedthroughout all the experiments, except in those where hydrogensulphide was used2556 NEWBERY : ELECTROMOTIVE FORCES IN ALCOHOL.PART V.Potential differences were measured by means of a carefullycalibrated metre potentiometer wire, a Weston cadmium cell beingused as standard, and a sensitive capillary electrometer as nullinstrument.Connexion of the potentiometer with the dropping mercury wasmade by means of a platinum wire dipping in the mercury in thepear-shaped vessel, and with the still mercury by a platinum wiredipping in the end of the syphon tube D.A is the leading tube of the experimental calomel electrode, andthe U-tube ‘below contained the same liquid as the calomel electrodevessel.With a pressure of 5 atmospheres, 250 C.C.of mercury fell intwentx-f our hours in N / 10-potassium chloride.(a) As a preliminary experiment, the absolute E.M.F. of thecalomel electrode in N/lO-potassium chloride solution a t 2 5 O wasdetermined. I n this experiment the following observations are ofinterest:(i) Rapid fluctuations of the potential of the still mercury wereobserved when the potential difference between the dropping andstill mercury exceeded 3 or 4 millivolts.These fluctuations wereobserved by Palmaer only when using hydrogen sulphide in forminghis null solution, and were attributed by him t o the presence ofsolid mercury sulphide. Similar fluctuations are, however, observedin many other cases, notably when the single potential of a metalcathode is measured during electrolysis of a dilute acid. The fluc-tuations are therefore in all probability due to the lack of balancebetween the solution pressure of the metal and the osmotic pressureof its ions present in the solution. They may be completely stoppedby the additjon of a soluble salt of the metal in question.(ii) The null solutions used by Palmaer a t 1 8 O were found to beunsuitable f o r use a t 25O, a greater proportion of potassium cyanidebeing required to render the solution null.I n the first experiment 40 C.C.of a solution containingO*lN-potassiurn chloride, O*OlN-potassium cyanide, 0.0008N-potass-ium hydroxide, and 0*0004N-mercury cyanide were placed in theelectrode vessel, and a similar solution without any mercury saltand with ten times the concentration of potassium cyanide wasplaced in the small burette.Henderson’s equation was employed in this and other cases t odetermine the diffusion potential due to dissimilarity of the liquidin the dropping electrode and calomel electrode vessels respectively.I n all caws it was found to be of the order 0.1 millivolt, and wastherefore neglected.The following table gives t,he last six readings, where column NEWBERY: ELECTROMOTIVE FORCES IN ALCOHOL. PART v.2557sliows the burette reading in c.c., I1 the E.U.P. of the calomelelectrode a,aainst the dropping mercury, and I11 the difference ofpotential between the dropping and still mercury :I. 11. 111.0.25 ,) 0.579 )) 0.6 7 ,0.29 ,) 0.582 ), 0.0 Y 70.14 C.C. 0-571 volt. 3.2 millivolts0.21 ,) 0.575 ,, 0.8 millivolt0.32 ,, 0.584 ,, -0.6 ,)0.48 ,, 0-611 ,, - 1.6 millivoltsHence the absolute potential of calomel electrode in N / 10-potass-ium chloride solution at 25" is 0.582 volt.This experiment was carried out three times, giving the values0.582, 0.582, and 0.583 volt respectively, the mean result being0.582 volt.Assuming the temperature-coefficient of the decinormal calomelelectrode to be 0.0008 volt per lo, this would lead to the value0-576 volt a t 1 8 O , as compared with Palmaer's value of 0.574 volt.( b ) Attempts t o determine the absolute potential of the calomelelectrode i n saturated alcoholic salt solution were made, the sameapparatus being used, but with the following modifications :(1) The calomel electrode vessel was specially made wit'h wideleading tubs and wide-bore tap (4 mm..).(2) Connexion between the two electrode vessels was made bybringing the liquids into actlual contact instead of using moistfilter paper, as was done with aqueous solutions.This was done byopening the pinch-tap a t C and sucking the air out of the joint.(3) A delicate Ayrton-Mather reflecting galvanometer was usedas null instrument inste'ad of the capillary electrometer.The sclution in the electrode vessel was 40 C.C.of a saturatedsolution of pure sodium chloride in pure alcohol (dried overcalcium), t o which was added mercury cyanide until 0'000125normal.The burebte contained a 0401N-solution of sodium cyanide inalcohol previously saturated with pure sodium chloride.The following table gives the last five readings, the figures I, 11,I11 having the same significance as before:I. 11. 111.0-14 C.C. 0.330 volt. 34.0 millivolts.0-32 ,, 0.330 ,, 21-4 ,;0.54 y y 0.331 )) 10-6 ,)0.68 ,, 0.331 )) 1.0 Y Y0.77 ,, 0-331 )) -33.4 ))Absolute potential of the calomel electrode in saturated alcoholicA second experiment by the same method gave the same resalt.VOL. cv.8 Dsodium chloride solution = 0.331 volt2558 NEWBERY : ELECTROMOTIVE FORCES IN ALCOHOL. PART v.A third experiment, using O*O'IN-sodium sulphide solution inplace of the sodium cyanide, gave 0.329 volt.A fourth experiment made six months later with a fresh set ofsolutions similar to those used in the third experiment gave again0.329 volt. Average of the four results=0'330 volt.( c ) An attempt was next made to determine the absolute poten-tial of the calomel electrode in saturated aqueous sodium chloridesolution.(i) A saturated solution of pure sodium chloride' containing0.000125N-mercury cyanide was placed in the electrode vessel, anda similar solution containing 0.13-sodium cyanide in place of themercury salt was put into the burette.On adding the solution from the' burette until the concentrationof the sodium cyanide was two hundred times that of the mercurycyanide, the potential difference between the falling and the stillmercury was reduced from 500 to 50 millivolts.Further additionof sodium cyanide produced a still smaller effect, so that it appearsto be impossible to produce a null solution by this method, whichwas theref ore abandoned.(ii) Palmaer's method of obtaining a null solution by passinggaseous hydrogen sulphide through the liquid was then tried, but,instead of adding acetic acid t o reduce the ionisation of the hydro-gen sulphide, it was found necessary to add sodium hydroxide t oincrease ii;.After making the saturated sodium chloride solutionnearly centinormal with respect t o sodium hydroxide and passinghydrogen sulphide for eight hours, the pot8ential difference betweenthe dropping and still mercury had only been reduced t o 10 milli-volts. The fig-ures obtained showed that if a null solution couldbe obtained by this method it would probably give a value ofmore thctu 0.6 volt for the absolute potential of the calomel elec-trode in saturated aqueous sodium chloride solution. Subsequentexperiment showed that this result is undoubtedly too high.(iii) Sodium sulphide was then tried as the agent for reducingthe Concentration of the mercury ions, a 0.001N-solution in satur-ated sodiurr, chloride solution being placed in the burette, whilst40 C.C.of a saturated sodium chloride solution containing0*000125N-mercury cyanide were placed in the electrode vessel.A slow current of pure hydrogen was passed through as in Z(a).The following table shows the results obtained, the figures I, 11,I11 having the same significance as before [Z(a)ii], whilst column IVshows the time interval between the readings:NEWBERY : EJ,ECTROMt)TIVE FORCES IN ALCOHOL. PART V. 2559I, 11. 111. IT'.0.00 C.C. 0-573 volt. 504 millivolts. 30 minutes0.72 ), 0.579 ,, 482 9 , 30 Y,1-44 ,, 0.586 ,, 277 ? ? 30 ,,2.25 ,, 0.595 ,, 206 9 , 21 hours2.25 ,, 0.591 ,, 255 ,, 30 minutes2-52 ,, 0.594 ,, 230 ,) 30 ,,2.70 ,, 0.595 ,, 174 ,, 30 ,,2-88 ,) 0.595 ,, 60 Y , 30 ),2-92 ,, 0.595 ,, -33.4 ,, 15 Y 72.92 ,) 0.593 ,, -10.6 ,, 21 hours2.92 ,, 0-593 ,, +238 9 ,1.80 ,, 0-588 ,, 250 1 , 30 ,,2-88 ,, 0.595 ,, 85 7, 15 ,,Frcin this table i t will be seen that a null solution is readilyobtained by the use of sodium sulphide, the potential of thedropping mercury remaining constant whilst the potential of the&ill mercury altered by 200 millivolts.The addition of sodiumsulphide only affected the potential of the dropping mercury to avery small extent when near the null point, although the effectwas slightly greater when far away from it.The effect on the potential of the still mercury was remarkable,although a t first it was comparatively small. When near the nullpoint, however, further addition of sodium sulphide, sufficient toincrease its concentration by 0*00005N, lowered the potential bymore than 100 millivolb, the greater part of which took placewithin the first ten minutes, reaching the maximum after aboutthree or four hours, and then falling again.Another remarkable feature of this experiment is the accuracyand certainty with which these results may be reproduced.Theexperiment was repeated three times, using fresh solutions eachtime, but exactly the same result (0.595 volt) was obtained. Byovershooting the mark with excess of sodium sulphide, and subse-quently returning to the null point by addition of mercurycyanide, the same result was again obtained. Also, after the seriesof experiments had been completed, the apparatus was dismantledand laid aside for six months.It was then refitted, new solutionswith materials from other sources used, and the above experimentrepeated, when exactly the same result was again obtained. Again,from the last line of the above table, i t will be seen that by allow-ing the liquid to remain (with the hydrogen passing a t the rate ofthree bubbles per minute) for twenty-one hours, the potential ofthe still mercury had risen again nearly 250 millivolts. On addingmore sodium sulphide the null point was again reached, and thefigure in the second column again became 0.595 volt. The actualquantity of sodium sulphide present thus appears to have little orno effect on the pocential so long as the null point is attained, andits addition can therefore cause no perceptible diffusion potential.8112560 NEWBERY : ELECTROMOTIVE FORCES IN ALCOHOT,.PART V.(iv) A similar experiment was carried outl with a normal solutionof sodium chloride in place' of the saturated solution. I n this casea greater concentratioii of sodium sulphide was required t o producethe null solution. The fluctuations of potential of the still mercurywere also greater and more persistent than in the previous case.Two determinations gave identical results, namely, 0.552 volt, asthe absolute potential of the calomel electrode in a normal solutionof sodium chloride in water.(v) When a similar experiment was tried with O*lX-sodiumchloride solution, the fluctuations of potential referred to previ-ously were very violent, a t times showing a variation of 40 milli-volts, and were never entirely absent.Also a high potentialE.U.F. appe'ared to be generated by the dropping mescupy, sincetouching any metallic connexion on the apparatus with the fingeror an earthzd wire a t once gave a large deflexion to the galvano-meter, which continued until the earthing body was removed. Asthe direction of this deflexion was determined by the side of thegalvanometer which was touched, i t was considered advisable toinsulate Lhe whole apparatus on blocks of paraffin-wax before takingfurther readings.Three determinations of the absolute potential of the calomelelectrode in O*lN-aqueous sodium chloride solution gave the values0.534, 0.535, and 0.535 volt respectively.(vi) It was found possible to produce a null solution in thecase of decinormal sodium chloride solations by the use of sodiunicyanide and also by the use of hydrogen sulphide and acetic acid,although neither of these methods could be used with the strongersolutions of sodium chloride.The use of sodium cyanide gave thefollowing result :Absolute potential of calomel electrode in N / 10-sodium chloridesolution = 0.579 volt.The use of hydrogen sulphide and acetic acid gave:Absolute potential of calomel electxode in AT/ 10-sodium chloridesolution = 0.570 volt.Summary.The following figures are given by the dropping electrode as theabsolute potentials of t'he calomel electrode in the solutions stated,a t 25O:(a) ( / I )In saturated sodium chloride (aqueous) ...0.595 volt.,, normal ,, ,, (aqueous) ... 0.552 ,, 7 Y 9 , 7 ) ,, (alcoholic) ... 0.329 ), 0.331 volt.y y decinormal ,, , Y 9 , ... 0.535 y y 0.579 y ,Y Y ,, potassium chloride ,, ... 0.572 ,) 0.582 ,NEWBERY: ELECTROMOTIVE FORCES IN ALCOHOL. PART V. 2561The figures in column (a) were obtained with the aid of analkaline sulphate, hhose in ( b ) with an alkaline cyanide.Of these figures, the value given in ( b ) for N/10-potassiumchloride solution at 2 5 O may be taken as approximately correct, asit is nearly the same as that) obtained by Palmaer (Zoc. cit.) whencorrected f o r temperature from the known coefficient of the calomelelectrode, and is also supported by Smith's observations with thecapillary electrometer (Zoc. cit.).The value given in ( b ) for N/lO-sodium chloride solution mayalso be taken as approximately correct, since it is within threemillivolts of that for potassium chloride.A concentration cell withcalomel ,electrodes and N / 10-solutions of sodium chloride andpot,assium chloride in opposition showed an E.M.F. of about2 millivolts.Even a superficial examination of the remaining figures will atonce show that they cannot represent true absolute potentials.The potential of the calomel electrode according t o these results isgreatest in the saturated solution, and least in the most dilute.This is directly contrary to theory, and will be shown later (PartVII) to be contrary to fact.Apparently the use of the alkali sulphide is responsible in parta t least f o r the abnormal results.It was made use of (1) becausei t was the' only substance found t h a t would give a null solutionat all with the stronger aqueous solutions, and (2) because inalcoholic solution i t gave almost the same result as sodium cyanide,which presumably gives true values at least' in aqueous solution.Again, the reason why the sodium sulphide produces a nullsolution with saturated aqueous sodium chloride, when hydrogensulphide will not, is by no means evident. If i t is suggested thatit is duel t o the greater extent, of ionisation of the sodium sulphide,then i t should follow that. more sodium sulphide should be requiredin the saturated sodium chloride solution than in the normal ordecinormal, f o r the sodium sulphide must obviously be less disso-ciated in t'he more concentrated sodium chloride solution. Thisconclusion is not supported by experimental facts, for it was foundthat the saturated sodium chloride solution required less than halfthe quantity of sodium sulphide that the normal solution required,and the normal solution required a little less than the decinormal.The foregoing data have been given at some length as theypossess so much of definiteness, and therefore must have somespecial significance which must be considered in any satisfactoryexplanation of the theory of the dropping electrode, for it will beshown in a future communication that these results, although s2562 JONES AND WHEELER :cleSiiite, afford no clue to the absolute potential of the calomelelzctrode in alcoholic or in concentrated aqueous solution.The author desires t o express his thanks to Professor Lapwortlifor his mcouragement during the progress of the work, and begst o acknowledge that some of the apparatus used had been pur-chased from a grant to Dr. Lapworth from tlie Government GrantResearch Fund of the Royal Society.CHEMICAL LABOltArOlLIES,MANCHESIER UNIVERSITY