EQUILIBRIA ACROSS A COPPER FERROCYANIDL ETC. 1313 CXXII. -Eyuilibria Acrom a Copper Ferrocyanide and an Amy1 Alcohol Membrane. By FREDERICK GEORGE DONNAN and WILLIAM EDWARD GARNER. DONNAN and Allmand (T. 1914 105 1941) investigatd the dis-tribution equilibrium of potassium and chlorine ions acro8s a copper ferrocyanide membrane. I n the method adopted a solution of potassium ferrocyanide was placed on one side of the membrane and a solution of potassium chloride on the other. The results, whilst affording undoubted confirmation of Donnan’s theory (Zeitsch. Elektrochem. 1911 17 572) were complicated by the uncertainty as to the manner of ionisation of potassium fern-cyanide. It was considered that this difficulty would be overcome if solutions of two ferrocyanides were employed on thO two sides of the mEmbrane especially if these salts were ionised to the same extent.I n the present investigation the sodium potassium and calcium salts were found to be suitable and mixtures of these were used to test the validity of the theory. With a mixture of sodium and potassium ferrocyanides equil-ibrium will be set up by an interchange of sodium and potassium ions since t-he membrane is not permeable to ferrocyanogen ions. Assuming that the ions obey the laws of ideal solutions the equa-tion for the equilibrium is given by where the symbols indicate molar ionic concentrations. This equation was derived from thermodynamical considerations by Donnan (Zoc. cit.) and its derivation is also possible from the kinetic theory as follows The number of ions of any one kind penetrating the membrane is proportional to its ionic concentration, C,; the number of ions exchanging across the membrane is pro-portional not only to the concentration C, but also to the concen-tration of the ion C2 which interchanges with the first across the membrane that is, N = K .C . C,. I n the case of sodium and potassium ions a t distribution equil-ibrium four kinds of interchanges across the membrane are possible. Two of these (a) and (6) (exchange of like ions) do not affect the final concentrations in any way. The four interchanges are shown below 1314 DONNAN AND QARNER EQUILIBRIA ACROSS A COPPER ( b ) [KO] +I- [K'] I A t equilibrium the number of exchanges due to ( c ) that is, N,=K[Na,']~,'] must equal that due to (d) that is, Nd = K[Na,'] [Klo] therefore [Na,'] [K,'] = [Naz'] [K,'].When the calcium salt is substituted for the potassium salt the equation of equilibrium becomes : ( a ) [Ca,'.] [Na,*l2= [Ca '][Na1'I2, (71) [Cal] [Na,I2 = [Ca,] [Na& and if the degree of ionisation of the two salts is the same. I n the case of the sodium-potassium cells which were investi-gahd the experimental results showed good agreement with the requirements of the theory. The degrees of ionisation of the two salts are very similar so that the ratio of thel molar concentrations of the salts is the same as that of the ionic concentrations. The sodium-calcium cells however gave unexpected results. Whereas the equation ( b ) which refers to the concentrations of the two salts holds within thO limit of the experimental error of the analysis it was found &hat equation ( a ) above does not accurately represent the relationship between the ionic concentrations of the calcium and sodium salts on the two sides of the membrane.The activities of the ions in this case appear to be more closely related to the molar than to the ionic concentrations. This result also may indicate that adsorption plays an important part in the trans-ference of the ions across the membrane. Further experiments were carried out in order to find a liquid membrane which would be permeable to one electrolyte and impermeable to a second which contains an ion common to the first. With this purpose in view the solubilities of several salts in moist organic solvents were determined.Amy1 alcohol was found to be the most satisfactory of these solvents and potassium and lithium chlorides the most suitable electrolytes. Since lithium chloride is readily soluble in amyl alcohol (a saturated solution is 1 -83N) and potassium chloride is practically insoluble (a saturate 1FICBBOCYANIDE AND AN AMYL ALCOHOL MEMBRANE. 1316 solution is 0*0048N),* it was hoped to set up a cell of the following I. KCI LiCl in water I Amy1 alcohol I LiCl in water 11. which is of the same type as that investigated by Donnan and Allmand with the copper f errocyanide membrane. It was not however practicable to use en amyl alcohol membrane owing to the slow rate of diffusion of lithium chloride through the amyl alcohol. The problem was therefore approached in an indirect manner.Determinations were made of the distri-bution concentrations of lithium chloride between amyl alcohol and water a t 2 5 O . Aqueous solutions of lithium chloride and lithium and potassium chlorides were shaken with amyl alcohol, and the two layers separated and analysed. The concentration of the lithium chloride in 11 which is in equilibrium with a mixture of the two chlorides in I was calculated from these results. The calculation of the ionic concentrations is complicated by the high values of the viscosity of the solutions and by changes in the state of hydration of the lithium ion with concentration. Green (T.? 1908 93 2023) has deduced the degree of ionisation of concentrated lithium chloride solutions from measuremenla of the conductivity of solutions of lithium chloride of which the viscosity has been increased by means of sucrose.The chief objw tion to the values which are obtained in this way lies in the hydr-ation of the lithium ion which will be affected by the addition of sucrose to the solutions. The concentrations of the ions and the undissociated part of the electrolyte have however been calculated using Green's values for the degree of ionisation of lithium chloride and the results are in fairly satisfactory agreement with theory. The agreement is better in those cases where the total concentration of the electro-lytes is below I>Tv'. [Li,'] . [Cl,/] = [Li,'] . [Cl,/]. The distribution-coefficient of lithium chloride between amyl alcohol and water has been calculated and it appears that lithium chloride occurs in amyl alcohol solution as double molecules.The coefficient is however only a constant over a small range of con-centration and above 5N the coefficient increases. The increase is probably associated with errors in the degree of ionisation due to the dehydration of the lithium ion. The experimental work in this paper leads to the conclusion that the same equilibrium relationships are established whether the * Moist &my1 alcohol. type : Thus [LiCI],=~iCl] and 3 D 1316 DONNAN AND GARNER EQUILIBRIA ACROSS A UOPPER equilibrium is brought about by the transference of ions as is the case with the copper ferrocyanide membrane or by the transference of the undissociated part of the electrolyte as is the cage with the amyl alcohol membrane.EXPERIMENTAL. The osmometer vessels used in the determination of the ratios of Che ions were those described by Donnan and Allmand (T. 1914, 106 1944). The copper ferrocyanide membrane in parchment paper was damped in position between two shallow cylindrical vessels and was separated from the supporting rim by rubber bands. The vessels which were of Jena glass were fitted with side-tubea t o facilitate the introduction of the solutions and the volume of each vessel was about 100 C.C. The membranes were prepared by the meth.od described by Donnan and Allmand (Zoc. cit.) and the parchment paper was usually left in contact with tbe solutions for two days. The membranes were tested for leaks by placing a ferrocyanide solution on one side and an isotonic solution of sucrose on the other and no leakage occurred over a period of six weeks.The ferrocyanides which were used in the investigation were purified by crystallisation from water. The calcium and ammonia ferrocyanides were prepared from hydrof errocyanic acid (No yes and Johnston J . Amer. Chem. Soc. 1909 31 991). Since potassium ferrocyanide forms insoluble double salts with magnesium and calcium ferrocyanides of the type it was not possible to use potassium ferrocyanide against these salts in the cells. The precipitation of the insoluble salts takes place slowly a t the ordinary temperature but quicker on heating, as if a chemical change were taking place. The double salts with sodium ferrocyanide are soluble in water.R,K2FeCy,,3H,O, Po tassium-Sodium Ferro cyanide Cells. The usual pro-cedure was t o place a solution of potassium ferrocyanide in one side of the cell and a solution of sodium ferrocyanide in the other. The time required for the attainment of equilibrium was deter-mined by conductivity measurements and no change in the con-ductivity could be observed after an interval of one week. The cells were however allowed to remain with occasional shaking, over a period of three t o five weeks in which time equilibrium was certain to have been reached. Solutions were used of a strength 0.025 molar FERBOCYANIDE AND AN AMYL ALCOHOL MEMBRANE. 1317 I n order to prevent changes in the ferrocyanide soluhioner with time several precautions were necessary-(1) When pieces of well-washed copper ferrocyanide membrane were placed in solutions of ferrocyanides it was observed that the strength of the solutions diminished several per cent.in three or four days and a t the same time the presence of sulphates in the solutions was detected. The change was almost entirely due t o that side of the membrane which was last exposed to the copper sulphate solution. The adsorbed copper sulphate on this side of the membrane reacted with the ferro-cyanide solutions with the forma-tion of sulphates and a slight in-crease in the thickness of the mem-brane. The copper f errocyanide membrane after being clamped in position was on this account washed for three t o four days with f errocyanide solutions of the same concentrations as those to be used in the experiment.(2) Another source of trouble was the oxidation of the ferro-cyanides by the small quantity of air enclosed in the osmometer vessels. The oxidation was also considerable if the conductivity of the solutions was measured from time to time in the ordin-ary conductivity vessels. To make the change due to this cause small as possible a special con-ductivity vessel was constructed (Fig. 1) which could be filled with nitrogen. The amount of air in the osmometer vessel was also reduced to a fraction of a c.c. and in the majority of the experiments of which the results are given in the tables the cells were not opened until immediately before analysis. (3) When solutions of different concentrations were employed on the two sides of the membrane osmosis of water was prevented by the addition to the solution of the requisite amount of sucrose.The amounts which were added were calculated from conductivity 3 Ds 1318 DONNAN AND GARNER EQUILIBRIA AOROSS A COPPER data. and Allmand have shown (Zoc. &.). two or three days. The effect due to this cause is however small as Donnan The solutions were kept in the dark and the celIs shaken every Method of Andysis. The solutions of the ferrocyanides were decomposed with con-centrated sulphuric acid and the sulphates of the alkali metals converted into chlorides by the precipitation of the iron (twice) with ammonia and of the sulphate with a slight excess of barium chloride. After the removal of the barium as carbonate the mixed chlorides in the solution were obtained by evaporisation and weighed.The potassium was determined as perchlorate and the sodium calculated by difference. The f errocyanide concentrations were determined before filling into the cells and after the equilibrium had been reached by titration against potassium permanganate solution. These analyses serve as a check on the result8 obtained by the gravi-metric analysis The variation in the ferrocyanide concentrations as determined by the three methods outlined above rarely exceeded 1 per cent. A method of analysis based on a conductivity method was not found to give the requisite degree of accuracy. The results for the potassium-sodium cells are given in table I. The weights of potassium chlorate and the mixed chlorides are given (in order to indicate the possible errors of the analyw) and in columns 6 and 7 are included the molar concentrations of the sodium and potassium on both sides of the membrane*A and B .In column 8 is found the total concentration of the metals and in 9 and 10 four times the total concentration of ferrocyanogen before and after the experiment. The results show that no large amount of oxidation or absorption of the salts has taken place during the period of the experiment. The ratios of the sodium and the potassium in the solutions on the two sides of the cell are compared in the last column and it will be observed that the ratio is the same for (a) and ( b ) within experimental error. The ratio of the ionic concentrations will be but little different from those given in the table since the degrees of ionisation of the sodium and potassium f errocyanides are very similar.The conductivity of 0.025 molar solutions of potassium sodium and ammonium ferrocyanides was determined a t 26O and the degree of ionisation calculated. The results are given in table 11 Cell No. 1 a .................. b .................. 2 a .................. b .................. 3 a .................. b .................. Pa .................. b .................. 5 a .................. b .................. TABLE I. Potassium and Sodium Ferrocyanide Weight8 of I ’F KCIO,. 0.3123 0.3146 0.2064 0.2069 0-1984 0.2181 0.1839 0.2163 0.1731 0.1785 pa. 0.2669 0.2666 0.1992 0.1982 0-1973 0.2181 0- 1970 0.2332 0.1936 0.1993 NaCl.0.0979 0-0965 0.0881 0.0869 0.0905 0-1007 0-0980 0.1168 0.1004 0-1033 KC1. 0.1680 0.1691 0.1111 0.1113 0.1068 0.1174 0.0990 0.1164 0-0931 0.0960 Na. 0.0336 0.0410 0.0603 0.0496 0-0516 0.0576 0.0569 00666 0.0673 0.0884 K. 0.0461 0-0667 0.0497 0.0498 0.0477 0.0624 0.0443 0.06206 0.0416 0.064 1320 DONNAN AND GARNER EQUILIBBIA AOROSS A COPPER TABLE 11. Degree of Zonisation of Ferrocyanides. salt. +40. rctusch. 1000. (NH.J,Fe(CN) ..................... 383.1 742.0 61.8 (Na),Fe(CN) ..................... 337.1 647.6 6% L (K),Fe(CN) ........................ 393*0* 742.0 62.9 A0 Kohl-* Noyes. Thus the relationship given below has been proved to be correct, that is, Since the activities of the potassium and sodium ions are prob-ably very similar the equation deduced by Donnan has been shown to hold.S o d ~ ~ m - ~ 4 m r n o n ~ t ~ m C e h . Ammonium ferrocyanide solutions slowly attack the copper ferrocyanide membrane. The membrane thickens and changes in colour from a dark brown to a reddish-brown It does not how-ever appear to break down as on on0 occasion a cell was made up of a solution of ammonium ferrocyanide on the one side and an isotonic sucrose solution on the other. After five weeks the sucrose solution was tested and it was found that no ferrocyanide had diffused through the membrane. The' concentration of ferro-cyanide had however diminished and the colour of the me,mbrane on the one side had changed to a reddish-brown.I n consequence of these irregularities only one sodium-ammonium cell was examined. The ratio ____ on the two sides were found to be 0.8480 and 0.8595 respectively. "H41 Sodium-Calcium Cells. The cells were made up as described previously. The amounts of sucrose used to prevent osmosis were the same as those used with the potassium and sodium solutions of the same concentra-tion. The concentrations of sucrose are probably too great in the case of the calcium solutions sinoe the calcium salt is less ionised in solution than the potassium salt but no appreciable osmosis occurred. On the other hand when the concentrations of sucrose were calculated from Sherrill's equations ( J . Amer.Chern. SOC., 1910 32 742) a considerable amount of osmosis occurred. The rate at which equilibrium is reached is about ita rapid a FBRROCYANIDE AND AN AMYL ALCOHOL MEMBRANE. 1321 in the case of the sodium-potassium cells. In some cases the changes in concentration were followed by conductivity measure ments. In Fig. 2 are given the changes of conductivity of a sodium-calcium and a sodium-magnesium cell respectively. Time indays. 0. 1. 3. 4. 5. 6. 7. 8. 10. 12. Bridge-reading. -Na-Mg . 2974 3358 3800 - 3994 - 4040 - - 4060 / -. Na-Ce . 3470 4082 - 4600 - 4635 - 4643 4635 -Time in days. From the curves and the above table it will be readily seen that the rate of exchange of sodium and calcium ions is approxim 1322 DONNAN AND UARNER EQUILIBRIA ACROSS A OOPPER ably equal to that of sodium and magnesium ions and that a constant reading is obtained in about ten days.The slight fall in the bridge-reading a t the end of the experi ment (Na-Ca) is probably due to the oxidation of the ferrocyanide, Method of Analysis. The analysis of the solutions of calcium and sodium ferro-cyanides gave rise to considerable trouble owing to the small volume (100 c.c.) which was available for analysis. The ferro-cyanide solution was evaporated to dryness and decomposed with concentrated sulphuric acid. The mixed sulphates were dissolved in dilute hydrochloric acid and the iron was removed as hydroxide with ammonia. The calcium was precipitated as oxalate and con-verted into oxide. Traces of iron were sometimes present in the oxide which was on this account dissolved in hydrochloric acid and the iron precipitated.The calcium was then weighed as sulphate. The filtrate containing sodium sulphate was evaporated to dryness traces of iron were removed and the sodium was weighed as sulphate. The results of the analyses are shown in table 111 and it is found that where the concentrations represent the total concentrations of the calcium and sodium atoms in the solution. The ratio of the equilibrium concentrations of the calcium ferro-cyanide on the two sides of the membrane is slightly higher than the ratio of the squares of the equilibrium concentrations of the sodium ferrocyanide but the variations are of the same order as those due to errors of analysis. The change in the degree of ionisation with concentration is known for potassium and calcium ferrocyanides but not for sodium f errocy anide.lO0u K,Fe(CN),. Ca,Fe(CN),. (1) 0.06 Mol- ............... 48.6 22.1 (2) 0.026 Molm ............ 63.1 23.5 (Noyes and Johnston J . Amer. Chem. Soc. 1909 31 1010). Similar figures have been obtaihed for calcium ferrocyanide in the course of this work. Assuming that the sodium salt resembles potassium f errocyanide, and that the ionisation of the mixed sodium and calcium salts is the game as in solutions of the pure salts with concentratioa corr TABLE 111. Calcium-Sodium Cells. Normality of 1 - r-4Fe(CN),. 2Ca. 00573 0.0453 00615 0.0349 0.0643 0.0397 0652 0.0362 0.1287 0832 Na. 0.0640 0.0573 0.0867 0.0654 0.0726 0.0629 0.0893 0.0670 0.0686 0.0661 I.01213 0.1026 0.1482 0.1003 0.1269 0.1026 0.1646 01043 0.1973 0.1393 Ir. 0.1196 0.1034 0-1495 0.1030 0.1286 0.1049 0.1634 0.1042 0.2140 0- 1303 111. 0.1185 0.1029 0.1473 0.1026 0.1023 0.1236 0.1645 0.1032 0.1995 0.1 420 CC%l rn 1.266 1.762 1.368 1.802 1.547 “%I’ “%I’ 1.248 1-767 1.332 1.776 1495 “bll C%I 1.117 1.326 1.1154 1.335 1.22 1324 DONNAN AND QARNER EQUILIBRIA ACROSS A COPPER sponding with the ferrocyanogen-ion concentration (Arrhenius) it is possible to calculate the ratios of the .calcium-ion concentration and the ratios of the squares of the sodium-ion concentrations on the two sides of the membrane.These are given in the table. The results (table IV) show that the relationship does not hold so strictly as equation (1). TABLE IV. Calcium-Sodium Cells. [ X I "all' [Ca;'I* "%'I: No. [Ca,I' "4 [Ca; 7 [Na2']*' 1 ..................... 1.27 1.25 1-25 1.19 2 ..................... 1.76 1.76 1.68 1.58 3 ..................... 1-37 1-33 1.33 1-25 4 ..................... 1-80 1.78 1.73 1.61 5 ..................... 1.65 1.50 1-49 1.36 The activities of the ions of these two salts thus appear to be more nearly proportional to the molar concentrations than to the ionic concentrations. Cell 3 which was opened once during the experiment shows that 2-3 per cent. of oxidation has taken place and in cell 5 a change due t o osmosis, occurred. Neither of these changes appears to affect the ratio to any great extent.I n cell 5 it should be noted that the concentration of the calcium ferrocyanide is greater than that of the sodium ferro-cyanide; in the other cells the reverse is the case. Irregularities occurred in two of the cells. Amy1 Alcohol Nembrane. The results of some preliminary experiments on this membrane are given below. Materials.-Amy1 alcohol (b. p. 131.5O) was obtained by repeated fractionation of fuse1 oil through a six-bulb fraction-ating column. One sample of the alcohol was used throughout the work. The lithium chloride was free from calcium and was completely soluble in amyl alcohol. Its solution in water was neither acid nor alkaline. It was also analysed by conversion into lithium sulphate followed by the estimation of the sulphate in this sub-stance as barium sulphate BERBOCYANIDE m D AN AMYL ALCOHOL MEMBRANE.1326 Method.-Aqueous solutions of lithium chloride were shaken with amyl alcohol in stoppered bottles in a thermostat kept a t 2 5 O , and when equilibrium had been reached the two layers were analysed. A dysis.-The aqueous solutions of lithium chloride were estimated volumetrically with N / 10-silver nitrate and the results were checked by analysis of the lithium as sulphate. A known volume of the amyl alcohol layer was placed in a Jena-glass distilling flask and the amyl alcohol distilled off. Water was added to the residue and the solution titrated with silver nitrate. The solutions containing the potassium and lithium chlorides were analysed according to the method employed by Gooch (Proc.Amer. Acad. 22 N.S. 14 177). A known volume of the aqueous solution was evaporated in the presence of 10 C.C. of amyl alcohol, and a little hydrochloric acid added to convert any lithium hydr-oxide into chloride. The lithium chloride dissolves in the amyl alcohol and potassium chloride is left behind. The residue is collected and washed with hot amyl alcohol. The lithium is then estimated as sulphate and the potassium chloride dissolved in water and estimated with silver nitrate. To check the results the solution of the mixed chlorides was titrated directly with silver nitrate. Results.-The concentrations of lithium chloride in the two layers are given in table V. The concentration of the lithium chloride in amyl alcohol diminishes rapidly with decrease in the concentration in the aqueous solution.The degree of ionisation of lithium chloride solutions cannot be given with any accuracy. The viscosity of the solutions is so great that allowance must be made in the derivation of the degree of ionisation from the conductivity results. Green (loc. cit.) has determined the TABLE V. Distribution of Lithium Chloride between Amy1 Alcohol and Water.* LiCIA [LialArn. [Licl]Aq. [LiclJAm. Total. Total. 100~. undissociated. [E 12.54N 8.49 7.77 6.68 5-00 3.14 3.00 2.7 1 1-735N 0.903 0.683 0.387 0.1266 0.0342 0-0314 0.0251 36.9 47.8 50.6 54-2 60.1 66.3 66.7 67.7 7,9137 4-43 3.85 3.06 2.00 1-06 1.00 0.88 0-0273 0.0428 0.041 8 0.0366 0.0277 0.0256 0.0262 0.0268 * In the calculation of the equilibrium constant a correction has been made or the d q p e of ionisation of the LiCl in the amyl dcohol solution 1326 DOHITAX ANXI QABNER EQUILIBRIA ACROSS A OOPPER viawity and conductivity of solutiona of lithium chloride ov0r a wide range of concentration and haa corrected for the effect of viscoeity by the addition of sucrose to the solution.The con-ductivity at infinite dilution was calculated over a wide range of viscosities ; the degree of ionisation is obtained directly from the A equation a=- where h is the conductivity of a solution of lithium nr chloride and & the conductivity a t infinite dilution of a solution of lithium chloride containing sucrose and with the same viscosity aa the first solution.Washburn ( J . Amer. Chem. SOC. 1911 33, 1461) finds that the relation between the degree of ionisation the conductivity and the viscosity is given by the relation a=. fo 4 l f ' 0 where f repreeents the fluidities and m=0'94 but this holds inly from 0-1'ON. For more concentrated solutions m varim with the concentration. In table VI m is given for the concentrations TABLE VI. Degree of Ionisation of Concentrated Lithium Chlam'de Solutions. Normality. 0.0 2.0 3-0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 f* 11 1-67 84-87 73-84 63.66 64.59 46.27 38.60 31-55 25.28 19.76 16.07 11.42 m. -0.900 0.858 0.832 0.812 0.798 0.786 0.771 0-769 0.749 0-738 0.729 A.115-3 61.63 52.58 44.76 ;I:; 26.04 21.05 16.725 13.225 10.395 8.149 loo [$)" 100a(Green). -68.4 66.2 61.9 68.7 55.6 62.0 48.4 44.8 42.0 39.6 37.2 -69.9 86.7 63.6 60.1 66.8 63.1 4945 46.0 43.1 40.6 38.2 2-12N and the value of rn decreases from 0.90 to 0.73. The degree of ionisation calculated from the above equation and the data given by Green will be found in columns 5 and 6. The values are the same as those obtained by Green t o within 1-2 per cent. The method of calculation is open to the objection that the lithium ion is probably hydrated in solution and that the conductivity at infinite dilution Aj is given by lithium ions which are prob-ably hydrated to a different extent than is the caw in a pure solution of lithium chloride with the same viscosity.The calcu-lated values of the degree of ionisation will therefore be the more accurate for the more dilute solutions. The values of a in table VI column 6 are used in tables V and VII. In table V the amount of undissociated lithiu FERBOOYANIDE AND AN SMYL BLCOHOL MICMBBANE. 1327 chloride (column 4) is calculated from these values of a. From the figurea in the last column it will be seen that lithium chloride is associated i n amyl alcohol solution to double molecules. A constant for [LicllAm- ~ [LiUlX,. is only obtained between 2N and 5N. Above this concentration the coefficient rises from 0.0277 to 0.0428, and then falls. The most probable cause of this deviation is dia-cussed above and it appears that the calculated concentrations of the undissociated molecules are too low.The molar concentrations of the solutions of mixed chlorides in I. (p. 1315) and the corresponding equilibrium concentrations of lithium chloride in the amyl alcohol are given in table VII. TABLE VII. Results with Amy1 Alcohol Membrane. Undissociated Total Total LiCl in Total LiCl [Li’] x [Cl’]. KC1. LiCl. amyl LiC1. - No. I. I. alcohol. 11. I. 11. I. 11. 1 0.944 3.504 0.554 3-78 1-44 1.36 5.94 646 2 1.200 2.613 0.303 2-95 0.945 0.976 4.046 3.900 3 0.962 5.45 0.2236 5-80 2.39 2-46 11.02 11-17 From the amount of salt dissolved by the amyl alcohol and the data in table V the corresponding values of lithium chloride in 11. are calculated and given in column 5.These figures represent the concentrations of lithium chloride in equilibrium with the solu-tion of mixed chlorides in columns 1 and 2 across the amyl alcohol membrane. The ionic concentrations are obtained from Green’s values for lithium chloride ‘and from Kohlrausch and Grotman’s values for potassium chloride. The degree of ionisation for the higher concentrations of potassium chloride are obtained by extra-polation from the latter values. The degrees of ionisation of lithium and potassium chlorides are apparently very similar. The ionic concentrations of the solutions of the mixed chlorides are calculated on the assumption made in the case of the calcium-sodium ferrocyanide cells. Cells 1 2 and 3 show good agreement with the equations [LiCl] = [LiCl] and [LiJ [Cl,’] = [Lii] [Cl,’]. The agreement, which is better than would be eqected supports the values for the degree of ionisation of lithium chloride which were obtained by Green. Summary. Determinations have been made of the equilibrium concentra-tione of solutions of sodium and potassium ferrocyanides an 1328 NIEEENSTEIN THE COLOURINQ MATTER OF (b) I. Sodium ferrocyanide ( c ) I. Sodium ferrocyanide Calcium f errocyanide 11. Ammonium ferrocyanide 11. The solutions in (a) ( b ) and ( c ) were in the neighbourhood of 0-025N. A liquid membrane has been investigated; amyl alcohol was chosen as the most suitable solvent and the electrolytes employed were potassium and lithium chlorides. (’) I* Lithium 1 Amy1 alcohol 1 Lithium chloride 11. Potassium chloride Lithium chloride gives rise to double molecules in amyl alcohol solution and a constant is obtained for the partition-coefficient up t o 5 N . So far as the preliminary experiments go the equilibrium con-centrations of the lithium and chlorine ions and the undissociated part of the electrolyte agree with Donnan’s theory. ~VERSITY COLLEQE, GOWER STREET, W.C. 1. [Received September 22nd 1919.