LIBRARY UNIVERSITY OF CALIFORNIA. RECEIVED BY EXCHANGE Class A Study of New Semipermeable Mem- branes Prepared by the Electro- lytic Method. DSSERTATION. SUBMITTED TO THE BOARD OF UNIVERSITY STUDIES OF THE JOHNS HOPKINS UNIVERSITY TN CONFORMITY WITH THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY. BY BENJAMIN F. CARVER, BALTIMORE, MD. 1903. EASTON, PA.: THE CHEMICAL PUBLISHING COMPANY 1903. A Study of New Semipermeable Mem branes Prepared by the Electro- lytic Method. DSSERTATION. SUBMITTED TO THE BOARD OF UNIVERSITY STUDIES OF JOHNS HOPKINS UNIVERSITY IN CONFORMITY WITH THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY. BENJAMIN F. CARVER, BALTIMORE, MD. 1903. EASTON, PA.: THE CHEMICAL PUBLISHING COMPANY 1903. CONTENTS. Page. Acknowledgment 4 Introduction 5 Previous Work on the Electrolytic Method. Work of Morse and Horn 5 Work of Morse and Frazer 6 Removal of Air from Cell Walls 8 Deposition of Membrane 10 Methods Used for Determining the Activity of the Membranes 12 Membranes Investigated 13 Phosphates. Calcium 13 Copper 14 Ferrocyanides. Cadmium 14 Zinc 16 Nickel 22 Cobalticyanides. Cobalt 28 Nickel 29 Copper 32 Ferrous 37 Sulphides. Cadmium 38 Conclusions 39 Biographical Sketch 40 186894 ACKNOWLEDGMENT. This investigation was suggested by Professor Morse and car- ried on under his personal supervision. It is with pleasure that the author takes this opportunity to thank Professor Morse for the assistance he has received from him, both in the carrying on of this investigation and in the laboratory throughout his Univer- sity work. The author further wishes to express his appreciation of the instruction and advice he has received from President Remsen, Professor Jones, Professor Mathews and Doctor Frazer. A STUDY OF NEW SEMIPERMEABLE MEM- BRANES PREPARED BY THE ELEC- TROLYTIC METHOD. INTRODUCTION. Since 1901, Professor Morse has been much interested in the electrolytic preparation of semi permeable membranes for the measurement of osmotic pressure. Up to this time no improve- ment has been made on the method of Pfeffer for the preparation of these septa. The method used by Pfeffer is beset with many difficulties and cells prepared by it generally give unsatisfactory results, as is proven by the fact that the very numerous attempts, which have been made to repeat Pfeffer' s works, have met with very little success. Pfeffer used membranes of copper ferrocyanide, calcium phos- phate and Prussian blue, but only obtained results of any value with the first of these. The highest concentration employed by him was a six per cent, solution of cane sugar, less than one-fifth normal. The unsatisfactory results obtained by the use of the Pfeffer cell and the difficulties encountered in its preparation by his method, led to an investigation, in this laboratory by Morse and Horn in 1901, on the preparation of semipermeable membranes by electrolysis. WORK OF MORSE AND HORN. 1 Results were obtained during this investigation which were very promising, and it was demonstrated that an active membrane could be deposited electrolytically with much more ease than by the Pfeffer method. At first ordinary battery cells were used and membranes of copper ferrocyanide were deposited in them, which showed osmotic activity. Cups of various kinds were then tested and it was found that active membranes could be deposited in all of them without difficulty. Bottle-shaped cells, made by a local potter, were then tried with a view of testing the power of elec- 1 Amer. Chem. Jour., 26-80. trolytically prepared membranes to withstand pressure. After depositing the membrane in these cells, they were filled with a normal solution of cane sugar, and closed with rubber stoppers through which glass tubes had been passed. They were then placed in beakers of distilled water, which stood on the floor, and the liquid rose in the tubes to the height of 5.5 meters (the height of the room), and then overflowed in periods ranging from six to fifteen hours. When the open manometers were replaced by closed ones containing mercury, the cells failed, not, however, in consequence of a rupture of the membranes, but as a result of the weakness of the porous walls of the cells. The next experiments they made were with small porous cups, such as are used in making standard battery cells. With these cells, pressures as high as 4.5 atmospheres were measured by closed manometers. Owing to difficulties which were encountered in securing the manometers in the cells, no higher pressures could at that time be measured. When pressur.es a little above 4.5 at- mospheres were reached, the sugar solution oozed out between the stoppers and the manometers and the stoppers with the manome- ters were forced out of the apparatus. The investigation of Morse and Horn showed, first, that mem- branes could be deposited electrolytically with little difficulty ; second, that a more effective means must be devised to hold the manometer in place and to prevent leaking between the stopper and manometer ; and third, that ordinary porous cells are not, as a rule, suitable for the deposition of membranes to be employed in the measurement of osmotic pressures, particularly high pressures. In 1902, Morse and Frazer continued the work of Morse and Horn and obtained some very interesting results. They found that porous cells made from coarse material were unsuitable for the measurement of high osmotic pressures, that the membrane deposited in such cells, also in others which were not hard burned was always located in the middle of the wall, thus leav- ing the inner part of the wall filled with water which dilutes the solution to an uncertain extent. With hard burned cells, which are less porous, the membranes were generally found to be depos- 1 Amer. Chem. Jour., 28-1. ited on the inner wall, which is a condition essential to accuracy in the measurement of osmotic pressure. They also devised a satisfactory means by which the manometer could be securely held in place and leaking at the stopper and manometer prevented. Cells made of very fine material and hard burned were obtained from the potter. Copper ferrocyanide membranes were deposited on the inner walls of these, and pressures as high as 14.5 atmos- pheres for half normal and 31.4 atmospheres for normal solutions of cane-sugar were measured. Although these high pressures were obtained in some cases, it was not an uncommon occurrence that the cells cracked or leaked at lower pressures, owing to the imperfect structure of the porous walls. Only about twenty per cent, of the cells, specially made for this work, withstood high pressure before developing weaknesses of one kind or another, and it became evident that cells of greater uniformity in respect to thickness, texture and strength were required for the economical prosecution of the work. Attempts to obtain cells of the right character from the potter having failed, an investigation of the conditions under which a suitable porous wall can be produced at will was taken up and is now in progress in this laboratory. The results of the investigation of Morse and Frazer demonstra- ted that the electrolytic method of depositing the copper ferro- cyanide membrane far surpasses the diffusion method of Pfeffer. But up to the time when the present work was begun, only that membrane had been deposited by the new process. The ease with which this membrane was deposited and the good results obtained with it, suggested that the method could be employed with advan- tage in the deposition of nearly every kind of precipitate which can be formed from electrolytes in solution, in which case it would afford a ready means of investigating a great variety of substances with respect to their character as semipermeable membranes. It was suggested that I try the applicability of the method to the deposition of a number of compounds which had been found to have or were thought likely to possess a semipermeable character. This paper contains the results of my investigation. The cells which were at hand when this investigation was un- 8 dertaken were of four varieties : First, the ordinary battery cells, with very porous walls of irregular thickness, which could be used to demonstrate but not to measure osmotic pressure. Second, a very thick- walled variety of small ones, which were very porous. Several of these were tried and gave very unsatisfactory results, owing, apparently, to the large size of the pores. Third, some few of the variety, out of which Morse and Frazer had selected the best for their use, but these were all found to be more or less defective in their structure and unsuited to the measurement of high pressures. Fourth, the bottle-shaped variety, with a capacity of about 200 cc., which had been used by Morse and Horn. From the experience of Morse and Frazer it was clear that none of these cells were suitable for the measurement of osmotic pressures, and that some other method must, for the time being, be employed to determine the activity of the membranes which it was proposed to investigate. Accordingly, it was decided to test the membranes, either by noting the rates at which the liquids rose in open mano- meters, or by ascertaining the rates at which the cells delivered their contents under a pressure only slightly above that of the atmosphere. THE REMOVAL OF AIR FROM THE CELL WALLS. The method used for the removal of air from the walls of the cell was the same, with a few unimportant modifications, as that used by Morse and Horn. The mouth of the cell, where the stopper was to come in contact with the porous wall, was covered with a thin coating of shellac in order to prevent the formation of a membrane in that place, which might easily be ruptured in re- moving the stopper, thus affording an opportunity for leakage when the cell was in operation. The cell was then closed with a rubber stopper, S, Fig. i, through which a glass tube, a, passed, the lower end of the tube being flush with the stopper. Two side tubes, b and c, were fused into the tube a, the tube b serving as an exit for the liquid as it rises in the cell, in consequence of the " endosmose," while the wire at- tached to the inner electrode g is passed through c. Through the tube a, the dropping funnel d was passed, the lower end of which reached nearly to the bottom of the cell. It was held in place by a piece of rubber tubing at h. The inner electrode g consisted of a small cylinder of platinum foil, to which a piece of platinum wire had been welded ; this wire passed up between the tubes a and d and out through the side tube c. The cell, thus arranged, was placed within a large platinum cylinder, F, which completely encircled it, and served as the outer electrode. The whole was then placed in a breaker, m, and the electrodes con- nected with the dynamo in such a manner that the current passed from f to g. It had been shown by Morse and Horn that the air 10 could be completely removed from the walls of a cell by means of the strong "endosmose " which manifests itself when a porous wall is made to separate a dilute solution of a salt into two portions, and a current passed from an electrode in one of these to an elec- trode immersed in the other. Hence a 0.05 per cent, solution of potassium sulphate was intro- duced into the cell, through the dropping funnel d, until the liquid reached the side tube b. The beaker .was also filled with the solution until the cell was completely covered. " Endosmose" appeared at once, and there was a rapid flow of the liquid from without, through the wall into the cell, sweeping the air with it. The overflow which escaped through the tube b was collected and measured. Generally a current was maintained, which gave an overflow of about 500 cc. per hour. From time to time, more of the solution was filled into the beaker to replace that which had passed through the walls of the cell. After about a liter had been forced through the walls in the manner described, the solution was partially replaced by distilled water, and the current continued until 200 cc. more of the liquid had been collected. Finally, the cell was removed, rinsed with distilled water and replaced. The beaker and cell were filled with distilled water and the electrolysis continued. This operation was repeated until a very high resistance was obtained, showing that the liquid was practi- cally free from the salt. The walls were now free from air and filled with nearly pure water, and the cell was ready for the depo- sition of the membrane. If it was not convenient to deposit the membrane immediately, the cell was filled with distilled water and placed in a beaker containing the same, where it was allowed to remain until required. DEPOSITION OF THE MEMBRANE. The method used to deposit the membrane was fundamentally the same as that used by Morse and Horn. The wet cell, with its walls filled with water, was fitted up in the same manner as it had been for the removal of the air, with the exception that the outer electrode, which was platinum in that case, was replaced by one of copper, zinc or nickel, etc., according to the composition of the membrane which was to be deposited. The electrodes were con- 1 1 nected with the dynamo or directly with the battery and the solu- tions filled in as nearly simultaneously as possible. In the beginning, owing to the fact that the porous walls were filled with nearly pure water, there was very little current. The current, however, within a very short time began to increase. Then, within a few minutes, when the membrane started to form, it began to fall, and continued thereafter to decrease steadily until the maximum resistance of the cell for the current employed was reached. This varied greatly. Ordinarily, it was above 800 ohms and in one case exceeded 14,000, while in another it was only 300 ohms. The resistance depends to a great extent on the size of the pores of the cell, the less porous giving the higher resistance, owing, no doubt, to the better support afforded to the membrane by walls having a closer texture. The battery, which was used in this work, consisted of fifty- five storage cells, divided into three sections, any one of which could be used separately or in combination with either or both of the others. With this arrangement a current with an electromo- tive force, ranging from 1 2 to no volts, was available. If a higher electromotive force was required, as was the case sometimes, the battery current was transformed by a one-horse power motor- generator to a 2 20- volt current. There was in series with the field of the dynamo a rheostat of so wide a range that it was pos- sible to depress the voltage by small intervals from its maximum, I. ., 220 tO 20 VOltS. The deposition of the membrane, in all cases, was begun with a i2-volt current, but when the resistance of this had increased to its maximum, a current with a higher electromotive force was employed, and so on until the current from the entire battery (no- volts) was in use, and, if this did not suffice, the battery current was transformed as described above. When the maxi- mum resistance was reached with this current, the deposition was discontinued. In some cases, however, the deposition was com- pleted with a lower voltage, owing to the fact that with a higher electromotive force, the current appeared at times to tear through the membrane causing the resistance to decrease rapidly. When- ever this occurred the voltage was depressed to that which had been in use previous to the break and the deposition completed with a current of lower electromotive force. 12 The formation of a membrane, that is, the time required to reach the highest resistance, usually occupied from an hour and a half to two hours. At first there was always some ' ' endosmose ' ' in the direction of the current, but this always decreased as the deposition of the membrane progressed. During the formation of the membrane it had been found nec- essary to replace the solution within the cell, from time to time, with a fresh one, in order to prevent the accumulation of alkali, which acts injuriously on the membranes. This difficulty was overcome very satisfactorily by filling the dropping funnel with the solution, and allowing it to run slowly into the cell during the deposition of the membrane. As stated in the description of the apparatus, the lower end of the dropping funnel was placed near the bottom of the cell, so that the fresh liquid entered below, pushing the alkaline solution above out of the cell continuously. Having completed the formation of the membrane, the cell was emptied, rinsed several times, within and without, with distilled water and then, if not immediately required, it was placed in dis- tilled water, filled with the same, and allowed to stand until it was needed for use. METHODS USED FOR DETERMINING THE ACTIVITY OF THE MEMBRANES. As stated before, the cells employed in this investigation were not suitable for the measurement of osmotic pressure with closed manometers. Therefore, the following two methods were used to determine the activity of the membranes : i. The cell, after the deposition of the membrane had been completed, was rinsed several times with distilled water and finally with small portions of the sugar solution (normal in all cases), which was to be used. The cell was then filled with the same, closed with a rubber stopper through which an open glass tube of small bore passed. It was then immersed in water and the height to which the liquid rose in the tube noted. As the liquid rose, other lengths of glass tubing were added until the ceiling of the room was reached. 13 2. The cell, after being treated as above, was closed with a stopper, through which one end of a glass tube, bent twice at right angles, was passed. The outer descending limb of the tube was cut off at a point a little above the level of the water in which the cell was immersed in order not to produce a siphon. The liquid, which was delivered at a pressure a little above that of the atmos- phere, was collected in measuring cylinders, and observations taken from time to time. Thus it was possible to determine the amount of liquid delivered in any given time. THE MEMBRANES INVESTIGATED. The membranes deposited and examined during this investiga- tion are, in the order in which they were tried, the phosphates of calcium and copper, the ferrocyanides of cadmium, zinc, and nickel, the sulphide of cadmium, and the cobalticyanides of cobalt, nickel, copper, and iron. The results obtained with these membranes are as follows : i. Calcium Phosphate. This compound had been employed by Pfeffer and was found to be semipermeable, but the results he obtained with it were not very satisfactory. In preparing the cells by his diffusion method, he used solutions of calcium chloride and of sodium phosphate. In this investigation tenth-normal solutions of calcium acetate and of sodium phosphate were employed. The membrane was first deposited in the walls of ordinary battery cells, the electrodes both being of platinum. When the cells, after the formation of the membrane, were partly filled with normal solutions of cane sugar, and immersed in distilled water, the level of the liquid within rose slightly, thus indicating some, but not very great, osmotic activ- ity. When the porous cups were broken, the membrane was found to have been deposited just within the inner wall. Its ap- pearance resembled somewhat that of wax. The phosphate was next deposited in the smaller variety of porous cups. These were filled with normal solutions of sugar, closed with rubber stoppers carrying open manometers, and placed in distilled water. The liquid rose in the tubes to heights ranging from one-half to two meters. The liquid in the tubes then became stationary for sev- eral hours and finally began to fall. As this behavior of the cell indicated that the porous wall did not afford the membrane suffi- cient support, it was suggested that an attempt be made to par- tially close the larger pores of the wall by introducing into them some crystalline precipitate before depositing the membrane. With this object in view, cells, from which the air had been pre- viously removed, were filled with a solution of sulphuric acid and surrounded by a solution of barium nitrate. The current was then passed for some time from the outer to the inner electrode, thus forcing the barium and sulphuric acid ions in opposite direc- tions through the walls and filling them with barium sulphate. The cells which had received this preliminary treatment were found to be little, if any, more effective than the others. It was, therefore, evident that no ad vantage was to be gained by plugging the larger pores in this manner before depositing the membrane, and the attempt was abandoned. The cells, when broken, showed that the barium sulphate had been deposited in one part of the wall, while the calcium phosphate was found in another, both deposits being continuous. Although the calcium phosphate membrane is easily deposited by electrolysis, and exhibits osmotic activity, it does not appear to possess this quality in the marked degree found in some of the compounds to be mentioned hereafter. 2. Copper Phosphate. Membranes of copper phosphate were deposited in cells of the same varieties as were used in examining the phosphate of cal- cium. Tenth-normal solutions of copper sulphate and of sodium phosphate were used in preparing the membranes. The inner electrode was of platinum, while the outer one was of copper, The membranes all showed osmotic activity, and the results ob- tained were, in general, about the same as those obtained with the calcium salt. The membranes, like those of calcium phos- phate, were found to be located just within the inner wall of the cells. The results obtained with the calcium and copper phos- phates, as well as the results obtained by others working in this laboratory with this class of salts, appear to warrant the conclu- sion, that the phosphates in general are not very active as semi- permeable membranes. 3. Cadmium Ferrocyanide. Tenth-normal solutions of cadmium sulphate and of potassium 15 ferrocyanide were employed in preparing this membrane. Both electrodes were of platinum. When a solution of cadmium sulphate is treated with the ferro- cyanide of potassium, a white amorphous precipitate of cadmium ferrocyanide is formed, which is dissolved by hydrochloric acid and, likewise, by potassium hydroxide. The behavior with the latter made it necessary to prevent the accumulation of alkali in the cell during the deposition of the membrane. The arrangement by which this was accomplished has already been explained. Resistances ranging from 800 to 1500 ohms were obtained while depositing the cadmium ferrocyanide membranes in the bottle- shaped cells, which were the only ones used in this and the sub- sequent work. CELL NO. I. The first cell prepared, as described above, was set up with a normal solution of sugar. An open tube about a meter and one- half in length was passed through the stopper closing the cell. The liquid rose to the top of the tube and began to overflow in a little less than five hours and continued to overflow for several days. The cell was then taken down, refilled with a fresh sugar solution, and several lengths of glass tubing were added to the original open manometer. After the liquid had reached the height of about two meters the membrane broke. Several other membranes of cadmium ferrocyanide were de- posited and although, in every case, in the beginning the liquid was forced upwards in the manometer at the rate of about two meters per hour, in none of them did it reach a height much over two meters before the membrane gave way. When these cells were rinsed with distilled water, before filling them with sugar solution, the water became clouded with small particles of precipitate, which were detached during the rinsing of the cell. When the cells were broken, the precipitate was found to be de- posited on the inner surface of the wall, and parts of the mem- brane could be readily detached. It is impossible to say at pres- ent whether a better membrane of cadmium ferrocyanide might not have been obtained under other conditions, that is, if the dep- osition had been effected with other concentrations of solutions, with a different electromotive force, or if better porous vessels had i6 been available. Nevertheless, in view of the fact that none of the other membranes behaved in the same manner, it appears probable that the cadmium ferrocyanide membrane is inferior to most of the others in respect to toughness and adhesiveness. 4. Zinc Ferrocyanide. This precipitate had been used by Tammann 1 in his optical method for the demonstration of relative osmotic pressures of so- lutions. The vessels used in this work were of the bottle-shaped variety. Tenth normal solutions of zinc sulphate and of potas- sium ferrocyanide were employed in preparing the membranes ; the electrodes were of platinum and of zinc, the former being placed in the potassium ferrocyanide solution within the cell, and the latter in the solution of zinc sulphate surrounding it. The white precipitate of zinc ferrocyanide, which is formed when the two solutions are brought together, was found to be insoluble in hydrochloric acid, but it was readily dissolved by potassium hy- droxide. It was, therefore, necessary, as in the previous case, to prevent the accumulation of alkali by frequently replacing the solu- tion within the cell by a fresh one, during the deposition of the membrane. The membrane in every case was deposited on the inner surface of the cell wall. CELI, NO. II. In the case of this cell about two hours were required to de- posite the membrane, that is, for it to reach its maximum resis- tance, which was 14,500 ohms. The voltage of the current at the end of the operation was 108, though in this, as in all the other depositions, a lower voltage was employed in the earlier stages. The cell, provided with an open manometer, was set up in the usual manner with a normal solution of cane sugar. The liquid began to rise in the tube at the rate of about two meters per hour, but when it had reached a point a little above two meters, it suddenly ceased to rise and began to fall, showing that the membrane had been ruptured in some part of the cell. The same unsatisfactory behavior of the cell was experienced several times during the course of this investigation. One cell would give good results from the beginning, while another of the same variety i Wied. Ann. 34. 299 (1888). 17 would prove to be almost a complete failure. .This difference in behavior was due to difference in the texture of the porous walls. Those which were of open texture and contained visible channels in the walls were of little service, while those which had a closer texture and were wholly free from visible holes gave invariably good results. It should be stated, however, that many cells of only moderate excellence could be considerably improved by re- peating the membrane-forming process after a rupture of the membrane. Usually the resistance of the repaired membrane considerably exceeded that of the original one. CELL NO. in. The second cell which was tried with a zinc ferrocyanide mem- brane gave much better results than the first, though the resist- ance of the membrane was much lower, reaching a maximum of only 3,750 ohms after one and one-half hours. The voltage of the final current was 96. The cell was set up with a normal solution of sugar. In closing it with the stopper the liquid was forced up in the tube to a height of o. 2 of a meter. The following table gives a few observations taken during the subsequent rise of the liquid in the manometer : TABLE No. I. Height of liquid in manometer. Time. Meters. 12.38 P.M b.2 12.52 " i.o 2.00 " ,V4 2-38 " ; 4-3 2.51 " 4-7 The tube, when the ceiling was reached, was bent over and the overflow collected in a graduated tube. During the first eighteen hours, 19 cc. were delivered. The cell continued to deliver for several days, the rate of delivery steadily decreasing, as the solu- tion within the cell became more dilute. The results obtained with these two cells afford a striking ex- ample of a case, where the membrane, which apparently offered the lower maximum resistance to the current during its deposition, gave better results when tested in respect to its osmotic activity. i8 It would, however, be unsafe to conclude from this and similar observations that the activity of a membrane is independent of its resistance to the current, as measured by the volt and ammeter, because we have no idea of the relative areas of the membranes in different cells. The areas of cell wall covered by two membranes may be equal in a given instance, while the effective areas of the two may differ greatly, owing to differences in the texture of the clay walls. Again, it should be noted, that the resistance, which a given membrane exhibits, is a measure only of the difficulty with which the current tears the membrane in certain parts ; and, therefore, a membrane of low resistance, if it has a relatively large effective area, may prove very active and satisfactory, judged by the amount of water which passed through it, provided the pres- sure to which the membrane is subjected is not sufficient to rup- ture it. CELL NO. IV. This cell was prepared in the same manner as Nos. II and III. The deposition of the membrane required two hours. The max- imum resistance obtained was 3,100 ohms with an electromotive force of 1 08 volts. The cell was set up in a manner different from that employed with the cells II and III. It was closed with a stopper carrying a small tube bent to two right angles, the free, outer, end of the tube being slightly higher than the top of the bottle, so that the contents of the cell might be delivered under a small but nearly constant pressure. The bottle was immersed to the neck in water at the temperature of the room, and the over- flow collected in a graduated tube. The observations made are given in Table II. The volume of the liquid in the filled cell was 180 cc. No at- tempt was made to maintain a uniform temperature, and the relative volumes of liquid delivered in equal intervals of time were, no doubt, considerably affected by the fluctuations of tem- perature, which amounted in some cases to i2C. Cell No. IV was taken down, rinsed with water and refilled with a fresh sugar solution. It was then placed in a bath, the temperature of which was 35C, and the overflow was collected as before. The observations made are given in Table III. TABLE II. Time. Delivery Time. Delivery in cc. in cc. 5.20 P.M. .. 10. .. ... 285.5 7-35 " 10. II. . ... 295.0 9.00 " .... 14.5 12... 307.0 9.30 " ... .... 16.0 I 3 ... 3I9-0 10.30 A.M..- 45.0 14... ... 329.0 I 1 91.0 15- . . . 338.0 128 o 16 . "M7 O 17 . . ... ?CA C 4- 182.0 18... . 360.5 6. 222.5 20. . . .. . 374.0 7 , 24.2 O 21 . . lSl.0 8 . . . ?S? O 9- 273.0 23- ' ... 392.0 Total time of delivery, 1167 hours. Total volume delivered, 460 cc. TABLE III. Time. Delivery in cc. Time. Delivery in cc. 4.00 ' 8.0 3.... 1.52.0 4-45 ' ' II. 4.... '75-0 IO.OO A. M 58.0 .5 .... >95-o I 3 . ' roo.o 6...- 209.0 *ime. Delivery in cc. 24 2 .... 404.0 25. 413.0 26. 419 o 27. 427.0 28. 433-0 29 44o.o 30. 445-0 31. 448.0 32. 45LO 33. 453-0 34- 456.0 35- 458.0 36. 460.0 Delivery Time. in cc. 7 224.0 8 235.0 9 246.0 10 255.0 II 263.0 Total time of delivery, 307 hours. Total volume delivered, 271 cc. The temperature of the bath, during the time of delivery, re- mained practically constant, not varying more than one degree. The results given in the table show a greater regularity in the decrease of the successive volumes delivered in equal intervals of time, than those in the preceeding table, where the temperature varied greatly from time to time. Cell No. IV was refilled for the third time with a normal solu- tion of sugar and replaced in the bath. The cylinders, in which the overflow was collected, were changed every twelve hours. At this period an attempt was made for the first time to dis- 1 Readings every twenty-four hours. 2 Readings every forty-eight hours. 8 Readings every twenty-four hours. 20 cover how the volume of liquid delivered in a given interval of time is related to the mean concentration of the contents of the cell during the same time ; in other words, to ascertain how the rate at which water passes through a semipermeable membrane is affected by the concentration of the solution. It is to be pre- sumed that, temperature and pressure remaining constant, the former is directly proportional to the latter, but there appears to be no experimental evidence bearing directly on the question. In order to ascertain the mean concentration of the contents of the cell during the successive intervals, the amount of sugar de- livered within each of them was determined by the method of Fehling. Knowing the amount of sugar present in the cell when it was set up, and the amount contained in the delivered liquid for each period, it is possible to calculate the mean concentration of the solution within the cell during any of the periods. Taking as an example, the period during which 18.8 cc. were delivered, the mean rate per hour was 1.57 cc. and the mean number of grammes of sugar present in the cell was 44.9. Dur- ing the previous period the mean number of grammes of sugar present in the cell was 50.0 and the mean rate per hour was 1.87 cc. Now if the volumes delivered during different periods are proportional to the mean concentration of the cell during those periods, then X in the following proportion should be equal to 1.57: 50.0 : 49.9 : : i : X. But X, or the calculated rate per hour, equals 1.67, and there- fore greater than that obtained or the rate of delivery during these two periods was not proportional to the mean concentration of the cell during the same periods. The observations made are given in Table IV. The results given in the table below do not agree as well as they should, but taking into account the fact that the sugar was determined by the use of Fehling' s solution, which is not a very accurate method, makes it evident that the unavoidable experi- mental error was quite large. The great differences in the re- sults at the end of the table are largely owing to the fact that some of the sugar solution had leaked from the cell and conse- quently the calculated concentration of the contents of the cell 21 was greater than it was in reality, hence, the calculated mean de- livery per hour was also greater than that obtained. When the cell was taken down the calculated amount of sugar which should have been present in the cell was 19.131 grammes. The amount found was only 9.072 grammes, showing a leakage of 10.059 grammes during the time of delivery. The error introduced into all the calculations by this leakage was small at first but increased with the time of delivery. TABLE IV. Column I in table Number cc. delivered in twelve-hour periods. Column II in table Number grams of sugar in cell at beginning of each period. Column III in table Mean number of grams of sugar in cell during de- livery. Column IV in table Calculated mean delivery in cc. per hour. I. II. III. IV. V. 61.596 29.0 53.03 57.31 2.41 22.5 46.971 50.00 1.87 2.10 18.8 42.834 44.90 1.57 18.3 39-295 4i.o6 1.52 16.8 36.313 37.8o 1.40 -67 .43 39 15-5 33.824 35-07 1.29 .29 13.8 3 T -8o7 32.82 1.15 .20 12.9 30-098 30.95 1-08 .08 1 1. 2 28.694 29.40 0.933 .02 ii. o 27.465 28.08 0.917 0.889 10.48 26.326 26.89 0.873 0.871 9.6 25.302 25.81 . 0.800 0.837 8.7 24.462 24.88 0.725 0.771 7.84 23.724 24.09 0.653 0.923 7.6 23.042 23.38 0.633 0.642 7.1 22.444 22.74 0.591 0.615 6.45 21.935 22.19 0.537 0.571 II. 22 1 21.134 21.53 0.466 0.516 9.60 20.667 20.90 0.400 0.452 6.80 20.242 20.40 0.283 0.390 Cell No. IV had been refilled with sugar solution three times without any repair of the membrane, that is, without repeating in it the membrane-forming process, but the results obtained in the successive experiments with it show that the activity of the 1 Readings every twenty-four hours. Capacity of cell 180 cc. 22 membrane was decreased very little, if any, by the large amount of water which passed through it. The amount of liquid delivered during each of the three periods and the time required for its delivery in each case are as follows : Time. Volumes delivered. Period i. 1167 hours 460.000. 2. 307 " 271.0 " 11 3. 348 " 276.8 ". Total, 1822 " 1007.8 " 5 . Nickel Ferrocya nide. Tenth-normal solutions of nickel sulphate and of potassium ferro- cyanide were employed in preparing the membranes. The porous vessels employed were of the usual bottle-shaped variety hitherto de- scribed. The electrodes were of platinum and of nickel, the for- mer being placed in the potassium ferrocyanide solution within the cell, and the latter in the nickel sulphate solution surround- ing it. The greenish white precipitate of nickel ferrocyanide, which is formed when a solution of a nickel salt is treated with one of ferrocyanide of potassium, is insoluble in hydrochloric acid, but when treated with caustic potash it is decomposed with the formation of nickel hydroxide. It was necessary, therefore, to exercise the usual care to prevent the accumulation of alkali within the cell during the deposition of the membrane. The membrane in all cases was deposited on the inner surface of the cell wall. CELL NO. V. In preparing the first cell, two hours and fifteen minutes were occupied in depositing the membrane. The final current had an electromotive force of 62 volts and the maximum resistance of- fered by the membrane was 1500 ohms. The cell was filled with a normal solution of sugar, closed in the usual manner and placed in a beaker containing about a liter of water at room tempera- ture. The method for testing the activity of the membrane was the same as that employed with cell IV. The observations made upon the delivery of the cell are given in table V. V. Time. Delivery in cc. Time. Delivery in cc. Time. Delivery in cc. 4-35 P-M 10 . . . 262 .0 26-. 409.5 5.03 " 2.O ii ..279.0 27... 414.5 7-4 " 14.0 12 . 287.0 28... 4I9.O 8.40 " 18-5 13 . . 298.0 29... 423.0 9-40 " 22.5 14 . . 309.0 30. ., 427-0 10.40 " ..'.-. 26.0 15 ..319.0 31... 431-0 IO.4O A.M 45-0 16 ..328.0 32 2 44O.O I 1 81.0 17 337-0 33--- 451-0 18 . . 346.0 34- 462.0 . T ^6 O . 161 o 20 ..365.0 36... 478.0 .... 5 182.0 21 373-0 37- 485.0 5 200 o 22 ..381.0 38..- 492.0 . 217 o 23 -.388.0 39- 498.0 8 A\ . . Total time of delivery 1241 hours. Total volume delivered 507 cc. The volume of the liquid in the filled cell was 200 cc. It con- tained, therefore, 68.44 grammes of sugar. When the cell, which was still delivering at the rate of 2 cc. per day, was taken down, the liquid within was found to contain only 7.12 grammes of sugar. In other words, the normal solution which originally filled the cell had been reduced to about one-tenth that concentra- tion in the course of the experiment. No attempt was made to maintain a constant temperature and the volumes, delivered during successive equal intervals, do not decrease with any degree of regularity. CELL NO. VI. A second cell was prepared in the same manner as the first. Two hours were required to reach the maximum resistance which was only 800 ohms. The electromotive force of the current, with which the deposition was completed, was 95 volts. The bottle was filled with a normal solution of sugar and set up in the same manner as No. V, no attempt being made to main- tain a constant temperature. 1 Readings every twenty- four hours. 2 Readings every forty-eight hours. 24 The observations made during the short time in which the cell was allowed to deliver are given in Table VI. TABI^E VI. Delivery Delivery Delivery Time. cc. Time. cc. Time cc. 4.15 P.M I 1 123.5 4 228.0 5-15 " 9-5 2 163.5 5 2 56.0 IO.OO A.M 7O.O 3 2OO.5 Total time of delivery, 138 hours. Total volume delivered, 256 cc. Having determined that the membrane was satisfactory, the cell was taken down, rinsed with water, refilled with a fresh nor- mal solution of sugar, and then placed in the bath, the tempera- ture of which was 35. The sugar contained in the delivered liquid was determined with Fehling's solution, as had been done in the case of cell No. IV, with the zinc ferrocyanide membrane. The results obtained are given in Table VII. TABLE VII. Column I in table Number of cc. delivered in twelve-hour periods. Column II in table Number of grams of sugar in cell at beginning of each period. Column III in table Mean number of grams of sugar in cell during de- livery. Column IV in table Mean delivery in cc. per hour. Column V in table Calculated mean delivery in cc. per hour. I. II. in. IV. V. 68.44 44.0 55-68 62.06 3.66 32.3 47.30 51-49 i 2.69 3-03 27.0 41.676 44-49 ' 2.25 2.32 23.1 37.643 39.6i .92 2.00 21-5 33.i6i 35-40 79 I.7I 19.2 30.431 31.80 .60 1.60 17.0 28.288 29.36 42 1.47 15.6 26.540 27.41 .30 1.32 1 Readings every twenty-four hours. 13.0 25-192 25-87 i. 08 1.23 12.7 23-83 24.53 i. 06 1.02 12.0 22.773 23.32 I.OO I.OO io.5 21.635 22.20 0.875 0-951 9.6 20.837 21.24 0.800 0.837 8-3 20.181 20.51 0.691 0.772 8.i 5 19.582 19.88 0.679 0.669 7-3 18.984 19.28 0.608 0.658 6.5 18.587 18.79 0.541 0.592 II.O 1 17.946 18.27 0.458 0.525 9-3 17.641 17.79 0.376 0-445 6-3 17-315 17.48 0.261 0.369 The results given in the table agree to about the same extent as those in Table IV. The same errors which entered into the calculation in that case were present also in this. The calculated amount of sugar which should have been present in the cell, when it was taken down, was 16.545 grammes, while the amount found was only 7.49, showing a leakage of 8.605 grammes during the time of the experiment. Cell No. VI was refilled for the third time with a normal solu- tion of sugar, closed with a stopper carrying a tube differing from those previously used in that the end of it reached nearly to the bottom of the cell. This method of procedure was followed in order to ascertain whether there was complete diffusion of the water as it entered the cell. There was reason for suspecting that the diffusion was incomplete (see Table XV) ; that a part of the water entering the cell through the membrane glided up the wall instead of diffusing uniformly, with the result that the liquid de- livered was, in general, less concentrated than the contents of the cell. With this arrangement, if the diffusion was complete, the concentration of the delivered solution should be equal to the mean concentration of the cell during the time of delivery, but, on the other hand, if the water as it entered the cell rose to the top, as was suspected, then the concentration of the delivered liquid should be greater than the mean concentration of the cell during the period of delivery, the pressure and temperature being constant. The cell was placed in the bath and observations made every twelve hours. 1 Readings every twenty-four hours. Capacity of cell 200 cc. 26 The concentration of the delivered liquid was calculated from its specific gravity, which was determined by a Mohr-Westphal balance. Only the first few observations made are given in Table VIII, because the sugar solution was found to have leaked through the membrane to such an extent as to render the latter observations worthless. TABI,E VIII. Capacity of cell 200 cc. Column I in table Number of cc. delivered in twelve-hour periods. Column II in table Number of grams sugar in cell at beginning of each period. Column III in table Mean number of grams of sugar in cell during period. Column IV in table Number of grams sugar in i cc.of delivered solution. Column V in table Mean number of grams sugar i cc. in cell during de- livery. Column VI in table Percentage relation of column IV and V. I. II. III. IV. V. VI. Per cent. uo. 44x1 44-7 54.297 61.368 0.3164 0.3068 I03.I 25.5 47-683 50.990 0.2594 0.2549 101.8 21.0 43.011 45.347 0.2225 0.2217 100.3 1948 39.078 41.044 O.2OI9 0.2052 98.3 1745 35-950 37.514 0.1793 0.1875 95-0 The amount of sugar found in the cell, when it was taken down, was 7.794 grammes while the calculated amount which should have been present was 22.04 grammes, showing a leakage of 14.246 grammes. The amount found in the water in which the cell had been immersed was 14.986. Since the error, which was introduced into the calculations in consequence of the leakage of the sugar solution through the membrane, tends to lower the percentage relation between the concentration of the delivered liquid and that calculated for the contents of the cell, it is safe to conclude that, even in the case of the last two observations, the concentration of the delivered liquid was in reality greater than the mean concentration of the contents of the cell during the time of delivery, in other words, that the water as it enters the cell does not diffuse rapidly enough 27 to give to the delivered solution a concentration equal to that of the contents of the cell. The present cell had been set up three times, without repairing the membrane. The amount of liquid delivered and the time oc- cupied in delivering it are as follows : Time. Volume delivered. Period i. 138 hours 256.0 cc. 2- 348 " 333.4 " 3. 252 " 263.3 " Total, 738 " 852.7 " PREPARATION OF POTASSIUM COBAI/TICYANIDE. Having obtained very satisfactory results with the ferrocyanide membranes it was decided to investigate some of the cobalticyan- ides. Since cobalticyanide of potassium in not a common labora- tory reagent, it was necessary to prepare it for the intended work. As text- books give rather meager directions for its preparation, it is not out of place to mention here a few precau- tions, which were found necessary in order to obtain satisfactory results. The salt was made by treating a solution of a cobalt salt (chlo- ride or nitrate, were both used in this investigation), with a solution of potassium cyanide until all the precipitate, which was formed in the beginning, was redissolved. The precipitate formed on adding the potassium cyanide solution is the protocyanide of cobalt, and this redissolves in the excess of potassium cyanide, forming the potassium cobalticyanide. The reactions which take place are : 2. 2Co(CN) 2 + 8KCN + 2H 2 K 6 Co 2 (CN) 12 + 2KOH -f 2H. The solution thus obtained was clear and had a yellowish brown color which became reddish brown on heating. The po- tassium cobalticyanide salt can be separated from its solution by one or two methods : First, by evaporation, when the salt sep- arates out in well formed crystals with a slightly yellowish color ; or second, by adding to the solution an excess of alcohol in w 7 hich the salt is insoluble. By the second method the salt is obtained as an almost white powder. Both methods were used. The salt obtained by the first was found to contain potassium cyanide and 28 a solution of it gave a strong alkaline reaction. When a solu- tion of cobalt sulphate was treated with it, a dirty pink precipi- tate was formed, which was dissolved by an excess of the rea- gent. Even after recrystallizing several times the cobalticyanide prepared by the first method, the product gave with cobalt sul- phate a brownish precipitate, but it was found that by adding acetic acid to the reagent until it became slightly acid to litmus paper, this difficulty disappeared and a clean precipitate of a rose-pink color was obtained. This was found to be a very satisfactory means for testing the potassium cobalticyanide solutions for alkali or potassium cyanide, before using them for the deposition of the membranes, since the presence of a very little alkali or cyanide can be detected by noting the color of the precipitate formed, when the reagent is added to a solution of cobalt sulphate. It is probable that the salt can be sufficiently freed from potassium hydroxide and cyanide by repeated recrystallizations, but the desired result is much more easily obtained by acidifying with acetic acid. Fur- thermore, since alkali is produced within the cell during the de- position of the membrane, it was found to be advantageous to add some acetic acid to the cobalticyanide solution, and this was done in all cases, whether the reagent had been prepared by the first or by the second method. The best results, however, were obtained by using cobalticyanide of potassium prepared by the first method, i. e,, by recrystallizing the salt twice from water and acidfying with acetic acid before using it for the deposition of the membranes. 6. Cobalt Cobalticyanide. Tenth-normal solutions of cobalt sulphate and of potassium cobalticyanide were employed in preparing the membrane. The electrodes were both of platinum, the negative within the cell and the positive without. When a solution of cobalt salt is treated with one of potassium cobalticyanide, a rather gelatinous, rose-pink colored precipitate of cobalt cobalticyanide is formed, which is insoluble in acids, hot and cold, but is readily decomposed by caustic potash with formation of the hydroxide of cobalt. This conduct with acaus- 29 tic alkali made it necessary to have acetic acid within the cell during the formation of the membrane. CELL NO. VII. Two hours were required for the deposition of the membrane, i. e., to obtain the maximum resistance, which was only 300 ohms. The electromotive force of the current at the close was 107 volts. The membrane was found deposited on the surface of the inner wall of the cell. The bottle was filled with a normal solution of sugar and closed in the usual manner. It was then immersed in water at the room temperature and the overflow collected in graduated tubes. The observations made are given in Table IX. TABUS IX. Delivery Deliverv Delivery Time. in cc. 1 in cc. Time. in cc . 10.00 A. M 31.0 12. . 184.0 24 223.0 I 1 53-5 TV 189.0 25 225.0 2 74-5 14.. 194-0 26 227.0 3 92.0 !5- 198.0 27 229.0 4 109.0 16.. 2OI.O 28 231.0 5 124.5 17.. 204.0 29 232.5 6 139-5 18.. 208.0 30 234.0 7 150.0 19.. 211. 31 235.0 8 160.0 20.. 213.5 32 2 . 239.0 9 169.0 21.. 216.0 33 242.0 10 i74-o 22.. 218.0 34 245.0 Total time of delivery, 781 hours. Total volume delivered, 245 cc. The results obtained show that the salt under investigation possesses a decided semipermeable character, however, the degree of activity in the one case tried was not as great as that observed is some of the other cobalticyanides which were examined. 7. Nickel Cobalticyanide. Tenth-normal solutions of nickel sulphate and of potassium cobalticyanide were employed in preparing the membranes. The inner or negative electrode was of platinum, and the outer or positive one of nickel. When a solution of a nickel salt is treated with one of potas- 1 Readings every twenty-four hours. 2 Readings every forty-eight hours. 30 sitim cobalticyanide, a voluminous, bluish green precipitate of nickel cobalticyanide is formed. This precipitate was found to be insoluble in acids, hot and cold, but when treated with caustic potash it was decomposed with formation of the pale green of hydroxide of nickel. Hence the usual precautions were taken in order to prevent the accumulation of alkali in the cell during the deposition of the membrane. CELIy NO. VIII. It required two hours and forty minutes to deposit the first membrane, and then the maximum resistance obtained was only 1 80 ohms. The final current had an electromotive force of 96 volts. The cell was filled with a normal solution of sugar and set up in the usual manner at room temperature. The observations made are given in Table X. TABLE X. Time. Delivery in cc. Time. Delivery in cc. Time. Delivery incc. 2.25 P.M. - - 14 3I5-0 30 478.0 7-40 " II. I 5 .... 327.0 3' 486.0 IO.OO A.M 45-o 16 ... 335-0 32 494.0 I 1 85.0 17.... 343.0 33 503.5 l8 .... ......352.0 34 512.5 7 . . 1 1.6 tj 19.... 360.0 35 519.5 O i O {J 'O 20 368.0 36 523-5 5 183.0 21 375-5 37 527.0 6 22 .... 383.0 38 530.5 2V* . , , . 7OC O 2Q . . . . r -1A O g 24 409.0 40 537-5 25 422.0 41'.... 544-5 jo 267 o 26.... 435-0 42 555-0 1 1 280 5 27.... 447-0 43 564-5 j 2 2Q2 ^ 28 458.0 44 571-9 29 469.0 45 577.5 1 O ' O U O' U 46 583-5 Total time of delivery, 1988 hours. Total volume delivered, 583.5 cc. The capacity of cell VIII was 210 cc. It contained in the be- ginning, therefore, 71.86 grams of sugar. When the cell which, 1 Readings every twenty-four hours. 2 Readings every forty-eight hours. 3 Readings every ninety-six hours. was still delivering as the rate of 1.5 cc. per day, was taken down, the liquid within was found to contain only 3.52 grams of sugar, that is, the normal solution which originally filled the cell had become diluted to about one- twentieth of its original concen- tration. CELL NO. IX. Cell No. IX was prepared in the same manner as No. VIII. It required two hours to deposit the membrane. The maximum resistance, which was 470 ohms, was obtained with a current having a voltage of 94. It was set up in the same manner as No. VIII, but was only allowed to deliver for a few days, when it was taken down, refilled with a fresh sugar solution, and placed in the bath. The observations made before removing it to the bath are given in Table XI. XI. Delivery Delivery Delivery Time. in cc. Time. in cc. Time. in cc. I. oo P.M. i 1 70.0 4 409.5 2.00 " 7.5 2 92.0 5 143.5 10.00 A. M 46.5 3 iio.o 6 158.0 Total time of delivery, 165.0 hours. Total volume delivered, 158.0 cc. Having determined that the membrane was satisfactory, the cell was, as stated above, refilled with a normal solution of sugar and placed in the bath which had a temperature of 35. The sugar in the delivered liquid was determined by the method of Fehling as had been done with cells Nos. IV and VI. Observations were made every twelve hours, and are given in Table XII. The results given in the table agree to about the same extent as those in Tables IV and VII. When the cell was taken down, the calculated amount of sugar which should have been present in the cell was 23.599 grams, while the amount actually found was only 9.985, showing a leak- age of 13.612 grams. Hence the calculations were vitiated by the same errors in this case as in the case of cells Nos. IV and VI. i Readings every 24 hours. 32 TABUS XII. Column I in table Number of cc. delivered in twelve-hour periods. Column II in table Number of grams of sugar in cell at beginning of each period. Column III in table Mean number of grams of sugar in cell during de- livery. Column IV in table Mean delivery in cc. per hour. Column V in table Calculated mean delivery in cc. per hour. v. i.8r 1.38 1.20 1. 10 1.0 5 1.003 0.912 0.895 0.747 0.761 0.888 0.825 0.719 0.648 0.552 o.SS 6 0.470 0.398 0.301 8. Copper Cobalticyanide. Tenth-normal solutions of copper sulphate and of potassium Cobalticyanide were employed in preparing the membranes. The inner or negative electrode was of platinum and the outer or positive one of copper. A turquoise-blue colored precipitate of copper Cobalticyanide is formed, when a solution of a copper salt is treated with one of potassium Cobalticyanide. This precipitate is insoluble in acids, hot and cold, but when treated with caustic potash, it turns green, and then becomes darker and darker in color until finally it has the black appearance characteristic of cupric oxide. 1 Readings every twenty-four hours. Capacity of cell 187 cc. I. 11. ni. IV. 63-99 24.7 56.42 60.7 2.05 18.2 5L5I7 53-97 1.51 15.6 47.633 49-58 1.30 14.2 44-518 46.08 1.18 13-5 41.740 43.13 1. 12 12.8 39-340 40.53 1.06 ii-5 37-421 38-38 0.958 IT. 2 35.7or 36.56 0.933 9-3 34.469 35-09 0.775 9-5 33.270 33.84 0.791 n. i 3I-9I9 32-59 0.925 10.3 30.665 31.29 0.858 8.9 29.674 30.11 0.741 8.0 28.830 29.25 0.666 7-75 28.112 28.47 0.645 I3-9 1 26.831 27.47 0-579 n.6 25-943 26.39 0.483 9-75 25.412 25.68 0.406 7-4 24.901 25.16 0.308 9.6 44-323 24.61 0.400 33 In order to prevent the formation of cupric oxide while de- positing the membrane, the solution of potassium cobalticyanide within the cell was acidified with acetic acid. CELL NO. X. In depositing the first membrane a maximum resistance of only 230 ohms was obtained at the end of two hours. The cur- rent, with which the deposition of the membrane was completed, had an electromotive force of 107 volts. The bottle was filled with a normal solution of sugar and set up in the usual manner at room temperature. The observations made, in respect to the overflow, are given in Table XIII. TABLE XIII. Time. Delivery Delivery in cc. Time. in cc. Time. 8. . . 27Q O T8 Delivery, in cc. 2 A* " . . AO7 O **<\3 4l6 O 147 O 12. .. 778 ^ 22 . I7A O T7.. . . 7/1 O 27.. .1/4.0 *,} ^q.y.u ,} 2o6 O 14 7^QO 24 ..271 "? 1^.. ..7600 2^.. . . 246 ^ 16. . . 778 ^ 26 . . 26l O 17 ^88 O 27 28. Total time of delivery, 885 hours. Total volume delivered, 513 cc. Cell No. X had a capacity of 217 cc. It, therefore, contained 74.25 grams of sugar when set up. When the cell, which was still delivering at the rate of 5 cc. per day, was taken down only 8.04 grams of sugar were found in its contents. In other words, the normal solution of sugar, which originally filled the cell, had been diluted to about one-ninth the original concentration. Cell No. X was taken down, rinsed with water, refilled with a fresh sugar solution and placed in the bath, the temperature of which was 35. The sugar present in the delivered liquid was determined by the 1 Readings every twenty-four hours. 2 Readings every forty-eight hours. 34 method of Fehling, as had been done in the case of cells Nos. IV, VI, and IX. The observations, which were made every twelve hours, are given in Table XIV. TABLE XIV. 1 Column I in table Number of cc. delivered in twelve-hour periods. Column II in table Number of grams sugar in cell at beginning of each period. Column III in table Mean number of grams of sugar in cell during de- livery. Column IV in table Mean delivery in cc. per hour. Column V in table Mean delivery in cc. per hour. Column VI in table Calculated mean delivery in cc. per hour. I. II. III. IV. V. VI. 74.257 35-8 36.35 63-865 69.06 35-4 3-03 27.65 56.408 60.14 35-2 2.30 j 2-63 24.60 50.861 53.63 35-8 2.05 ! 2.05 22.50 46.327 48.59 35-4 1.88 -85 20.75 42.537 44-43 35-4 1-73 -72 19-43 39423 40.98 35-8 1-54 59 18.32 36.659 38.04 36.0 1.53 .38 17.40 34.516 35-59 35-8 1-45 43 15.35 32.975 33.78 27.0 1.28 37 13.40 31.343 32.16 22.0 1. 12 .22 10.45 30.130 30.64 2 4 .0 0.87 .06 14.13 28.610 29-37 35-o 1.18 c >-833 11.30 27.431 28.02 32.0 0.941 i .130 12.55 26.198 26.81 36.0 1.045 < ).8 9 8 10.60 25.182 25.69 32.0 0.883 ] [.03 II. 12 24.147 24.66 34-0 0.926 o .844 9-50 23.330 23-74 29.0 0.791 o .891 During the latter half of the period in which the cell was in operation, the temperature of the bath fell several times to that of the room in consequence of accidents to the gas supply during the night. This will account for the great irregularities of tempera- ture given in the table. The temperatures given are those of the bath, when the measuring tubes were changed. The results obtained are of little value, except as they indicate the activity of the membrane and in a rough way the effect of temperature on the amount of liquid delivered. The amount of sugar which, according to calculations, should 1 Capacity of cell, 217 cc. 35 have been in the cell, when it was taken down, was 23.33 grams, while the amount found was 19.53, showing aleakage of 3.8 grams during the experiment. Cell No. X was refilled for the third time with a fresh sugar solution and replaced in the bath. The sugar contained in the delivered liquid was calculated from the specific gravity of the latter which was determined with a Mohr-Westphal balance. The object was to ascertain what relation existed between the concen- tration of the overflow and the mean concentration of the contents of the cell during the period in which the liquid was delivered. If there was complete diffusion, of the entering water, within the cell, then the two concentrations should be equal, pressure and temperature being constant. Only a few of the earlier observa- tions are given in Table XV, however, because the error intro- duced into the calculations in consequence of leakage of the sugar solution through the membrane, became so large during the latter part of the experiment that the results obtained were of uncertain value. XV. 1 Column I in table Number of cc. delivered in twelve-hour periods. Column II in table Number of grams of sugar in cell at beginning of each period. Column III in table Mean number of grams of sugar in cell during per- iod of delivery. Column IV in table Number grains of sugar in i cc. of delivered solu- tion. Column V in table Mean number of grams of sugar in I cc. in cell dur- ing delivery. Column VI in table Percentage relations of columns IV and V. I. II. III. IV. V. VI. Per cent. 36.1 63-991 69.124 0.2844 0.3185 89.2 2 7 .82 57.3io 60.650 0.24015 0.2794 85-9 2 3-75 52.154 54-732 0.21711 0.2522 86.0 22.73 47.636 49.895 0.1988 0.2299 86.4 20.92 43-773 45-704 0.1799 0.2106 85-4 19.42 40.501 42.137 0.1685 0.1941 86.8 i6. 3 8 37-934 38.717 0.1576 0.1784 87.7 17.02 35.486 36.710 o. 1436 0.1691 84-3 Capacity of cell 217 cc. 36 The results make it appear that the concentration of the delivered liquid was always considerably less than that of the con- tents of the cell during the time of delivery and the difference is too great to be accounted for by any errors in the calculations due to leakage. It appears, therefore, that the water as it enters is not immediately diffused through the contents of the cell. The same conclusions was reached in the case of cell No. VI. When the cell was taken down, the calculated amount of sugar within was 20.66 grams, but the amount found in the contents of the cell was only 15.47, showing that a leakage of 5.19 grams had taken place during the experiment. Cell No. X had been rilled three times with fresh sugar solu- tion. The volumes delivered and time required for their delivery are as follows : Time. Volumes delivered. Period i. 885 hours. 513.00 cc. " 2. 204 " 295.40 " 11 3- 252 " 331-81 " Total, 1341 hours. 1140.21 cc. These results show that the effectiveness of the membrane in this cell was decreased very little, if any, by the large quantity of water which passed through it. CEU< NO. XI. Cell No. XI was prepared in the same manner as cell No. X. Two hours were occupied in reaching the maximum resistance, which was 2,750 ohms. The deposition of the membrane was completed with a current, having a voltage of no. The cell, when set up in the usual manner, started to overflow so slowly that it was immediately taken down and the membrane-forming process repeated for a period of forty-five minutes. At the end of this time a maximum resistance of 5,550 ohms was obtained with a current having a voltage of 112. The cell was refilled with normal sugar solution and again set up at room temperature. The observations made are given in Table XVI. 37 XVI. Delivery Delivery Delivery Time. in cc. Time. in cc. Time. in cc. I2 -55 P.M .......... - 3 .......... 109.0 9 .......... 175.0 J -55 " .......... 6.0 4 .......... 123.0 10 .......... 181.0 10.30 A. M .......... 44.0 5 .......... 136.0 ii .......... 186.0 i 1 ................ 70-0 6 .......... 147-0 12 .......... 191.0 2 ................. 9 1 - 7 .......... 163.0 13..-'. ...... 196.0 8 .......... 169.0 14 .......... aoi.o Total time of delivery 351.0 hours. Total volume delivered 201.0 cc. p. Ferrous Cobalticyanide. Tenth-normal solutions of ferrous sulphate and of potassium cobalticyanide were employed in preparing the membrane. The electrodes were both of platinum, the negative within the cell and the positive without. When a solution of ferrous sulphate is treated with one of potassium cobalticyanide, a slightly yellow, amorphous precipitate of ferrous cobalticyanide is formed. This precipitate is somewhat affected by strong acids, especially nitric, but acetic acid did not appear to change it. When treated with potassium hydroxide, a mixture of ferrous and ferric hydroxides was formed. It was necessary, therefore, to prevent an accumulation of alkali in the cell during the deposition of the membrane. CELL NO. XII. The membrane was deposited in the usual manner. Two hours and forty minutes were occupied in obtaining the maximum resistance, which was only 600 ohms. The voltage of the current, with which the deposition was completed, was 107. The cell was filled with a normal solution of sugar and closed in the usual manner. It was placed in a beaker containing about one liter of water at the room temperature. The overflow from the cell was collected in graduated tubes. The observations made are given in Table XVII. 1 Readings every twenty-four hours. 38 TABLE XVII. Time. Delivery ] in cc. Time. Deliver in cc. 10.00 A.M. 19.0 II 173-5 T Q T c 2 . . 39- I2 ' 101.5 3 78.0 14 199.0 O2 5 T^ 5 101.5 16 214.0 6 114.0 17 220.5 7 124.0 18 227.0 8 Q . I