LIBRARY UNIVERSITY OF CALIFORNIA. RECEIVED BY EXCHANGE Class A Study of Some New Semi- permeable Membranes. DISSERTATION SUBMITTED TO THE BOARD OF UNIVERSITY STUDIES OF THE JOHNS HOPKINS UNIVERSITY IN CONFORMITY WITH THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY. BY J. P. COONY, S. J BALTIMORE, 1903, A Study of Some New Semi- permeable Membranes. DISSERTATION SUBMITTED TO THE BOARD OF UNIVERSITY STUDIES OF THE JOHNS HOPKINS UNIVERSITY IN CONFORMITY WITH THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY. BY J. P. COONY, S. J BALTIMORE, 1903. CONTENTS. PAGB. Acknowledgment 4 Introduction 5 Historical Sketch 5 New Membranes by Electrolysis 5 Preparation of the Cell 6 Apparatus 7 Electrical Equipment 8 The Membrane : Its Position 10 Formation 11 Activity and Resistance 11 Methods of Testing 12 Ferric Hydroxide 12 Ferric Phosphate 16 Manganese Ferrocyanide 17 Deterioration 19 Cobalt Ferrocyanide 20 Influence of Temperature on its Activity 20 Prussian Blue 23 General Conclusions 25 Leakage 27 The Cell Wall 27 Relations between Concentration and Rate of Flow 28 Comparison Curves 31 Residual Concentration ..... 32 Perfecting the Wall 34 Addendum 36 Biography 37 186895 ACKNOWLEDGMENT. To Professor Morse, without whose initiative, kindly assistance and experienced direction this work could not have been carried out, the author wishes to make grateful acknowledgment. No less is he indebted to President Kemsen, whose lectures have been both guidance and inspiration. The able instruction of Professor Jones, Pro- fessor Shattuck and Professor Clark, likewise calls for appreciative recognition. A STUDY OF SOME NEW SEMIPEKMEABLE MEMBRANES. It is quite unnecessary, in giving an account of work done on the subject which forms the title of this paper, to review even briefly, the work and measurements of Pfeffer, 1 the brilliant theoretical work of van 't Hoff 2 founded on them, or the extremely valuable results achieved by De Vries, 3 Tammann, 4 and Hamburger, 5 men who were stimu- lated to undertake their great work by van 'tHoff's con- clusions. The great importance of the subject has made the scientific world familiar with these results. And if subsequent progress has failed to correspond to these bril- liant beginnings, the explanation is to be sought for in the almost insurmountable difficulties everywhere to be met with. The failure of so many attempts to repeat Pfeffer's work demonstrated the fact that only by great good fortune could an experimenter hope to secure a good membrane and a reliable cell by the methods employed by this investi- gator. Indeed, no work comparable to his had been done until a new method, devised by Morse 6 and Horn, and further perfected by Morse 7 and Frazer, gave well founded hopes of easily and surely forming quite perfect mem- branes. The results attained by these last workers are most interesting indeed, as they show the great rapidity with which a satisfactory membrane develops a pressure sufficient to burst very highly resisting materials. Near Osmotische Untersuchung (1877) . Zeitsch. phys. Chem. I. 481 (1887). Ibid. 2, 415 (1888). Wied. Ann. 34, 299 (1888). Zeitsch. phys. Chem. 6, 319 (1890V Amer. Chem. Journ.26, 80 (1901). Ibid. 28, 1 (1902). 6 the end of the paper last mentioned we read the brief his- tory of a cell in which the measurement of the osmotic pressure of a normal solution of sugar was attempted. " Resistance of the membrane not exactly known, but over 200,000 ohms. The solution of sugar fresh and free from invert sugar. Cell set up at 5 P. M. Pressure at 5.10 P. M., 7.89 atmospheres; at 6.45 P. M., 31.41 atmospheres (corrected). Temperature 24. 9. Shortly after the second reading the glass tube was shattered." The very satisfactory results recorded in this last paper gave a new impetus to work in this field and the prospect of soon being able to measure directly and with accuracy the osmotic pressure of normal solutions, and probably of vet higher concentrations, made clear the immediate desirability of a number of semipermeable membranes of a sufficiently wide range of chemical characteristics to permit of the selection, in each individual case, of some one membrane which should be chemically inert with regard to the par- ticular substance whose osmotic pressure measurements were desired. For the discovery of such membranes the electrolytic method seemed to promise speedy and sure re- sults, and it was accordingly made use of in the following investigation. In testing the various semipermeable membranes investi- gated in this laboratory, two cups have been employed. The smaller kind has a capacity of about 15 cc. and seems in general to possess a finer texture and to have been burned to a greater degree of hardness than the larger variety, the capacity of which varies from 170 to 200 cc. This larger cell, because of its shape, is commonly called the bottle cell. THE PRELIMINARY PREPARATION OF THE CELL. This preliminary treatment consists first in melting shellac within the neck of the cell until a firm coating is secured, covering all that portion of the cell wall which could be affected by the introduction and removal of the rubber stoppers employed. The purpose of this film of shellac is to make sure that no portion of the semiper- meable membrane formed will be exposed to the danger of rupture by friction or by the strains set up in the neck of the cell by the tightly fitting stoppers. By this means is precluded all possibility of a leakage of the solutions through portions of the porous wall where either no mem- brane has been formed, or, if formed, has been ruptured or locally removed in the process of closing the cell. The next step is to free the cell wall of all soluble mate- rials, especially of the air contained within the pores. As the simple apparatus here employed is used also in the formation of the membrane, it will be well to describe it in detail. A rubber stopper through which passes the direct arm of a glass T tube of say 10 mm. internal diameter is fitted into the neck of the cell. This admits a second glass tube of three or four mm. diameter. This second tube, the purpose of which is the transmission of liquids into the cell, reaches quite to the bottom of the cell, and is the outlet from a reservoir of suitable capacity. The flow from this into the cell can be regulated at will by a pinch cock on the rubber connection. Surrounding the lower end of this second tube is the inner platinum electrode, the platinum connection of which passes also through the T tube. The horizontal arm of the latter serves as an escape for the liquids introduced into the cell from the reservoir. By this means the liquid within the cell can be renewed as desired without any interruption to the process. In most cases frequent renewal is advisable. The cell is then placed in a beaker, of a depth sufficient to permit of complete immersion, and is surrounded by the outer electrode. In the process of washing cells, and frequently also in building up the membrane, this electrode is of platinum, though in the latter process, other workers in this laboratory have 8 found electrodes of Cu, Ni, Sn, Zn, etc., to be more satis- factory when the membrane to be deposited was a salt of one of these metals. When the membrane is some insolu- ble ferric salt, both electrodes must be of platinum, as a test made with an iron anode showed that by the electrolytic action an appreciable portion of ferrous compound was formed together with the ferric salt. The cell walls should first be thoroughly wetted from either the outside or the inside. This is done in order to avoid the probable entrapping of air between the advancing walls of liquid should the latter be applied to both sides of the cell wall at about the same time. When thus wet- ted, both the cell and the beaker in which it stands are filled with a 0.05% solution of potassium sulphate in recently boiled distilled water. With such apparatus the process of washing is easilv carried out. The inner electrode is o J made the cathode and the current intensity so regulated that the endosmose amounts to 8 or 10 cc. per minute. The outer solution needs replenishing at times to supply loss and to keep the cell covered, and both outer and inner solutions require occasional renewal, as they shortly become weak in cations and anions respectively. Two hours of this treatment is followed by a similar treatment with boiled distilled water. The current is passed until a very low conductivity shows the nearly complete absence of potassium sulphate. This leaves the cell walls quite free of foreign materials, and the membrane may now be introduced, or the cell may be kept in distilled water until the operator is ready to proceed to this step. THE FORMATION OF THE SEMIPERMEABLE MEMBRANE. The Electrical Equipment. The current employed is usually taken from the storage batteries whose combina- tions permit the use of any potential between 2 and 110 volts. Connections are easily established, also, with two dynamo circuits of 110 and 220 volts respectively; while three rotary transformers furnish currents at any desired potential below 300 volts. In the large and coarser cells low voltages are always found necessary in the earlier stages of the process; frequently 12, and at times even 6 volts marking the upper limit of the current which may safely be employed. In such cases, the use of a stronger current seems to prevent any rise in the electrical resist- ance of the membrane. It is thought that under such conditions the membrane is unable to bridge over many of the pores in the cell wall, which, by a less vigorous treatment might be effectually closed. Thus a large and un- varying cross-section of the liquid is left for the unimpeded passage of the ions. The safer, and, indeed, more expedi- tious method seems to be to confine the current to limits of one, or at most two-tenths of an ampere, raising the volt- age somewhat as the resistance of the membrane rises; though an exception seems to be necessary in the case of the Prussian Blue membrane. With this membrane no good results were obtained until either a small current for a long period, or a large current for a shorter time had almost choked the cell walls with the amount of the pre- cipitate formed. The Solutions. The sulphates or nitrates of the metals were preferably employed. It is evident that when plati- num anodes are used the chlorides are necessarily barred. Concentrations. Though varying conditions have made changes in concentration advisable, the practice of Morse and Horn and of Morse and Frazer of generally using N/10 solutions has been found to give the best results. Theoretically, this question of concentration should be determined by a consideration of the relative dissociations of the electrolytes at the concentrations employed, as also of the relative ionic velocities of the two constituents of the membrane. But the practical solution is much more easily carried out, and consists in simply observing the position of the membrane whether within, or upon either edge of the porous wall and in changing the relative concentra- tions accordingly. While no systematic study has been made of the advan- tages or disadvantages of the various positions in which the membrane may be deposited, it is obvious that a membrane formed on the outer edge of the wall can have no support whatever against pressures from within. The fact is, that cells in which the membrane has been deposited upon the inner edge of the wall have been found to be uniformly much more active than those in which the membrane was formed within the wall. This was to be expected, as diffu- sion takes place comparatively slowly within a porous body of close texture and considerable density ; and instead of having the semipermeable membrane as a fine partition sharply separating pure water from a solution of say normal concentration, the incoming water would soon so dilute the solution in the pores of the wall that the concentration at the membrame would probably fall far below one-tenth normal, and the activity, as indicated by the rate of flow of the water through the membrane, would be correspondingly low. If, on the contrary, the semipermeable membrane is on the inner edge of the wall, the practically undiminished concentration of the solution within, and the pure water circulating through the porous wall, are separated only by the thin membrane, and the maximum flow will be observed. In the matter of measuring osmotic pressures also and for the same reason a maximum pressure could hardly be expected, even after a very long time, if the membrane be within the wall. In consequence of these considerations, efforts have been made to so correlate the concentrations of the two electro- lytes that the semipermeable membrane should form upon the inner edge of the wall; and as experiment has shown 11 that by using high relative concentrations within, the pre- cipitate can be formed even in the outer electrolyte, and vice versa, the matter of position is under complete con- trol. In practice, a beginning is made with concentrations probably correct, regard having been had for ionic veloci- ties and dissociations of the electrolytes, and, if the mem- brane forms where it is desired, as is commonly the case, no change is made. Otherwise, one or the other solution is diluted accordingly. TO FORM THE SEMIPEEMEABLE MEMBRANE. When the cell has been prepared and the apparatus set up as before described, connection is made with the proper electrical terminals, and the electrolytes are in- troduced nearly simultaneously. The electrical resistance falls during the short period required for the electrolytes to displace the water from the pores of the wall and rises as soon as the formation of the membrane begins. Under o proper conditions, the resistance continues to increase until a maximum has been reached which corresponds to the character of the porous wall and the capabilities of the particular membrane. These observations coincide with conclusions reached by Morse and Horn, who, working with cups of considerable porosity, found that the resist- ance of the membrane could not be raised above a certain point. This maximum seems to be determined by the number and size of those pores which the membrane is unable to bridge. If, after a delay of some days, the membrane-forming process is repeated, a considerably higher maximum re- sistance is usually obtained. This second maximum is seldom exceeded in later repetitions, even though these be quite numerous. If, however, the concentrations be so changed as to form the membrane in some new position, the resistance becomes quite irregular. Such irregularity 12 was very marked in the earlier work. Instances also occurred of the resistance falling considerably below the maximum by the mere continuance of the process. This phenomenon is thus far without a satisfactory explana- tion. TESTING THE MEMBRANE. After the membrane has been formed the cell is washed thoroughly with distilled water, filled with the desired solution usually a sugar solution exactly normal and closed with a tightly fitting rubber stopper containing either a manometer or a simple delivery tube. It is then piuced in a vessel containing sufficient distilled water to rise above the highest level reached by the membrane. The mem- brane may be tested for activity, that is, for the rate of flow of the pure solvent through the membrane into the solu- tion within ; or, it may be tested for the pressure developed with a given concentration. In the former case the delivery tube is so arranged that its lowest point is about 10 mm. above the level of the outer liquid. The cell is, in conse- quence, always operating against this slight pressure. If it is desirable to test the membrane for pressure and this was done only in the early part of the work the rubber stopper contains a closed manometer in which the com- pression of a volume of carefully purified air indicates the pressure developed. THE VARIOUS MEMBRANES. The Ferric Hydroxide Membrane. The properties of this well-known precipitate, ferric hydroxide, seem such as to make it suitable in the highest degree for a semiperme- able membrane. Its insolubility, its firm gelatinous con- sistency, are almost too well known to all those who have attempted the gravimetric determination of iron. The difficulty experienced in washing the precipitated hydroxide 13 at once suggests the idea of impermeability with regard to dissolved materials. Its chemical inertness with regard to most basic substances strongly recommend it for the meas- urement of the osmotic pressure of this class of compounds. As the only ferric salt immediately obtainable was the chloride, which is manifestly unsuitable for use with plati- num electrodes, the first attempt was made with ferrous sulphate, giving a membrane of ferrous hydroxide. The expansion of this upon oxidation to the ferric hydroxide, would, it was thought, produce an extremely close semi- permeable membrane. A bottle cell of 185 cc. capacity was filled with a N/15 solution of ferrous sulphate, to which was added a considerable amount of carefully washed ferric hydroxide precipitate, introduced for the purpose of taking up the sulphuric acid set free by the electrolysis. Outside, a N/30 solution of sodium hydrox- ide was employed. This lower concentration was thought sufficient because of the much higher velocity of the hydroxyl ions. The membrane began forming almost im- mediately, and with a potential of S3 volts the resistance of the cell increased quite regularly at the rate of nearly 20 ohms per minute, reaching 825 ohms in forty-five min- utes. The rise in resistance was less rapid after this, and over an hour was required to reach the maximum of 1000 ohms. The resistance an hour later marked 800 ohms. The cell was next thoroughly washed with distilled water, then immersed in the same liquid, while through the interior air bubbles were continuously forced during eighteen hours. How complete was the oxidation was never known. The membrane, however, never exhibited any osmotic activity, and attention was next turned to- wards the preparation of pure ferric salts; the sulphate, the nitrate and the acetate. When these were obtained cell II of the same kind was filled with a N/2 ferric sul- phate solution, while the outer liquid was a N/12 sodium 14 hydroxide solution. Here the resistance when the mem- brane first began to form was 12 ohms. In 45 minutes it rose to 385 ohms. The cell was then allowed to stand over night in distilled water, the formation of the mem- brane being resumed on the following morning. The resistance now rose rapidly, going beyond 1000 ohms within thirty -five minutes. At this point the cell was washed, and then set up with a normal sugar solution, but without any manifestation of activity. The formation of a successful membrane in this cell was next attempted from ferric nitrate. The resistance of the mem- brane reached a maximum of 2680 ohms one hour after start- ing, and in the next fifteen minutes dropped to 2300 ohms. Here again, the cell was washed, and again set up with the normal sugar solution, but without a more favorable result than before. In the third attempt, the concentration of the sodium hydroxide solution was made equal to that of the ferric nitrate within. This gave a low resistance varying not far from 100 ohms, and the fast moving hydroxyl ions, passing entirely through the wall, formed an abundant precipitate of loose ferric hydroxide inside the cell. A last attempt was made employing ferric acetate. The highest resistance obtained was 175 ohms. The cell showed no osmotic activity with either a normal sugar solution or a twice normal solution of sodium chloride. One week later the outer liquid was intensely salt; from which we infer that the ferric hydroxide membrane, in such a wall is completely permeable with regard to this substance. Cell XVIII. This cell, of the same grade, was tried six weeks later when considerable experience had been gained with other membranes in the same type of cell. Three days were spent in futile efforts to form an active membrane of ferric hydroxide. The cell when set up with a normal sugar solution showed no activity, and twenty- 15 four hours later the marked sweetness of the outer solu- tion proved either that this kind of cell does not properly support a ferric hydroxide membrane, or that this mem- brane is permeable as well towards sugar as towards sodium chloride. With the small cells of finer texture the ultimate result was widely different. In two hours with a N/10 solution of ferric acetate within and a N/20 solution of sodium hydroxide without, the resistance rose to 9000 ohms the maximum resistance developed in this cell. This was, comparatively speaking, a lower figure than those obtained in the larger cells ; regard being had for the correspond- ingly large cross-section of conductive membrane. Yet, when set up with a N/2 sugar solution, the liquid rose rapidly to the top of the open manometer tube. This had a length of 1.2 m. and a bore of about 1 mm. The rate of rise of the liquid was about 10 mm. per minute. This cell continued to flow freely during the next three days. It was then taken down and fitted for measuring higher pressures. The membrane was reinforced by a further electrolytic deposition of ferric hydroxide, and the cell, after having been filled with a normal sugar solution, was fitted with a closed manometer and placed in the distilled water. The highest pressure recorded was 0.5 atmos- pheres. This seemed to indicatete that the insufficiently supported membrane gave way locally, allowing the leak- age at this pressure to equal the inflow of water The known activity of the membrane and the amount of sugar passing into the outer water could not otherwise be ac- counted for. A second cell of this type, which had failed to produce an active membrane by a similar treatment with ferric phos- phate, was tried with the ferric hydroxide membrane. This cell when a resistance of 3000 ohms had been reached, was set up with an open manometer tube 2.4 m. 16 in height. The level of liquid in this tube rose at the very rapid rate of 92 rnm. per minute when at the height of 600 mm., and the sugar solution flowed freely from the top of the tube during the two days it was allowed thus to remain. Four other cells of the same type were tried with this membrane, the results being uniformly satisfactory as regards activity, but never showing a pres- sure much in excess of 1 atmosphere. Of these cells the flow of one was measured. The capacity of the cell was 16 cc. and it was delivering at a pressure of about 35 mm. water. The figures record the total delivery after twenty- four hour periods. First 24 hours 8.0 cc. Second " 13.2 " Third " 16.2 " Fourth " 18.4 " These cells were afterwards broken and their membranes carefully examined. They appeared firmer and more uni- form than the other membranes tested in this investigation. There is every probability that in a porous wall of a suffi- ciently fine and uniform texture this membrane will prove to be of great value. The cells thus far tried with it are known to give unreliable support under pressure : indeed, the cells of the coarser type failed, even when no pres- sure was developed ; and hence we may reasonably ascribe to them and not to the membrane the deficiencies mani- fested. The Ferric Phosphate Membrane. A fair trial of this membrane cannot, it is probable, be made until a cell wall of the desired texture is forthcoming. Two cells of the finer grade were given careful trials, yet they failed to indicate any osmotic action whatever. These trials were not suffici- ently exhaustive to prove the complete incapacity of these cells for supporting the membrane, yet it is clear that the 17 membrane could not be satisfactory in them. The appear- ance of the ferric phosphate precipitate is most promising, and its chemical character such as to make it extremely valuable if an active membrane can be formed from it. The Manganese Ferrocyanide Membrane. This precip- itate is of fine grain with very little tendency toward ag- glutination, and it does not offer the appearance of a prom- ising semipermeable membrane. The color, white at first, changes on standing to a delicate, pale green. This change occurs as well in the cell when filled with a sugar solution as in any vessel in which the precipitate maybe formed. In de- positing this membrane it has always been found necessary to employ considerably lower voltages than with other membranes under similar conditions. The highest poten- tial which could be safely used in the first application of the membrane-forming process to a cell of the coarser variety was 12 volts, nor was any higher voltage deemed safe when, after a four days' test with a normal sugar solution, a reinforcement of the membrane was found necessary. In the third treatment of this cell, four weeks later, any at- tempt to go beyond 33 volts at once lowered the resistance. This fall of resistance continued until a return was made to the lower voltage. This membrane was tried in cells of both types, audits behavior was, in general, unsatisfactory. Its activity as manifested by the rate of flow under a pressure of 10 mm. of water was about one -half that of the ferric hydroxide membrane, about one-third that of the cobalt ferrocyanide membrane, and two-thirds that of the Prussian Blue mem- brane under similar conditions. This comparison in the case of the ferric hydroxide membrane could be made only in the smaller cells ; in the other cases the larger cells were employed. The first cell tried was one of the larger variety. The solutions were N/10 manganese sulphate and potassium 2 18 ferrocyanide, and the current from the battery with a potential of 12 volts was passed continuously during two hours and forty -five minutes ; a maximum resistance of 345 ohms being reached at the end of two hours of the electro- lytic action. After washing, the cell was set up with a normal sugar solution to deliver at a pressure of 10 mm. ; the temperature being about 15. The flow at first was at the rate of over 5 cc. per hour, but this activity soon diminished, reaching, in three and one-half days, the small figure of 9.1 cc. per day, with a total delivery of 65 cc. Here the membrane-forming process was repeated, a resist- ance of 1167 ohms being reached in one hour. At first the rate of delivery was slower than before, but the staying qualities of the firmer membrane were plainly manifest. Two weeks later with a total delivery of 177.5 cc. the rate was still 9 cc. per day; though frofn this time on the diminution was marked, only 3.7 cc. being delivered in the same period ten days later; the total delivery in twenty-four days being 233.4 cc. The third attempt to build up a firm membrane in this cell showed no improvement over the second. The resist- ance was nearly the same, the activity of the membrane somewhat less than before, and the total amount delivered in thirty-seven days only 224.5 cc. A fourth and fifth time similar efforts were made to reinforce this membrane. The activity as indicated by the rate of its measured de- livery seemed somewhat diminished each time, while the resistance remained practically the same. A peculiarity of this cell was that after it had been set up for two or three days, the sugar solution showed a slightly yellowish tint and contained appreciable, and constantly increasing amounts, of very fine white suspended matter, not distin- guishable from the magnanese ferrocyanide. In sixteen days this turbid condition was quite marked both in the water surrounding the cell and in the solution within. The latter 19 solution had become quite opaque. This characteristic was persistent throughout the whole history of the cell, not only in its numerous trials with sugar solutions but also on being tested with alcohol. Here the delivery was -of a very marked yellow color. In other cells the yellowish tint was somewhat in evi- dence, but not the turbid character of the liquids within and without. Cell XXII. of the same kind and with the same membrane, duplicated in every respect, except the last, the results just recorded. Two tests with cells of the finer grade showed nothing more than that the membrane was better supported in these cells and that it continued longer in them without deterioration. In one of these cells the curve representing the rate of delivery as plotted against concentration showed no appreciable deterioration of the membrane during thirty-two days ; and when, thirty days later, the cell was broken for the inspection of the membrane the latter was quite equal in appearance to the sound and firm ferric hydroxide membrane. With this single exception, all the ferrocyanide membranes tested showed some dete- rioration. This was best observed when the cell, after hav- ing been subjected to the desired number of tests, was broken, and the membrane compared in appearance with a newly formed membrane of the same type. It was also perceived by a falling off in the delivery of these cells in the later periods of their history. This deterioration con- sisted mainly in a partial transformation of the ferrocyanide into the corresponding oxide, though complete local removal of the membrane also took place. In the manganese ferrocyanide membrane the deterioration was chiefly of the latter kind and was very marked, being greater than in any other membrane. In the cobalt ferrocyanide membrane both forms of deterioration took place, but in a lesser degree ; while in the Prussian Blue membrane only local removal seems to have occurred, and this to no great extent in the very thick membranes. 20 The Cobalt Ferrocyanide Membrane. The next mem- brane studied was that precipitated from cobalt sulphate and potassium ferrocyanide. This precipitate is flocculent, and quite cohesive. When freshly precipitated, it is green- ish in tint, changing to a bluish-green, and, in very old membranes, to a purple color. There has never been any difficulty in forming a membrane from this precipitate. With N/10 solutions of these salts and a battery current of 12.4 volts, the resistance of the membrane in one of the larger cells rose steadily, reaching 910 ohms at the end of two hours and thirty minutes. When set up with a normal sugar solution the temperature being that of the laboratory and averaging between 15 and 16 the membrane showed an activity notably greater than any hitherto observed. Without having been reinforced, this membrane continued to manifest the same high degree of activity throughout the entire period of its first test. At the end of fifty-five days the delivery amounted to 4.4 cc. per day, the total delivery had been 574 cc., and the concentration of the sugar solution within had dropped to N/61. With no further treatment than washing with distilled water, this cell was refilled with a fresh normal sugar solution and placed in a constant temperature bath at 35. The fact that in seven days it delivered 256.7 cc. as compared with 226.2 cc. for the corresponding period of the first test shows that its activity had not been seriously impaired. It also confirms observa- tions made earlier in the history of this cell, viz., that the activity is notably greater at higher temperatures. This work was done in early January, and although the labora- tory temperature approached 20 in the afternoon, a some- what lower temperature was the rule during the night and the morning hours, and the rate of delivery as indicated by the readings, and yet more strikingly by the time elapsing between the falling of successive drops from the delivery 21 tube, followed these temperature variations quite closely. The following figures copied from the notes taken at the time will perhaps be of interest. They cover the fourth, fifth, and sixth days of this test. Time. Amt. Delivered. Temperature. No. of Seconds. 850A.M. 148.4 cc. 228 10.30 " 150.5 12.30 P. M. 153.6 4.45 " 159.7 10.00 A. M. 176.8 2.00 P. M. 181.5 4.00 " 184.0 10.45 A. M. 201.0 12.15 P. M. 202.0 2.45 " 204.8 14. . 209 17.3 197 19.7 185.5 13.0 271.5 16.3 257 19.7 220 15.0 284 16.o 275 18.2 265 Similar readings were taken covering a period of ten days, indicating the same dependence of rate upon tem- perature. Observations taken at the same time on the action of other cells show a repetition of these variations. This membrane was not, however, possessed of indefinite endurance; and, after ten days of the second trial, gave evident signs of weakening. It was accordingly reinforced. A battery current of 61.5 volts was employed during a period of two hours, at the end of which time the steadily growing resistance registered 2010 ohms. A second rein- forcement two weeks later developed a maximum resist- ance of 2040 ohms, the current employed being taken from the dynamo circuit at 103 volts. In both instances the cell fully equaled its former record for activity. In a second cell of the same type this membrane de- veloped a resistance of 1150 ohms in one hour and thirty minutes, the battery current being taken at 38 volts. When set up in a constant temperature bath at 35 the delivery was very rapid, 9.9 cc. in one hour, 109.6 cc. in one day, and 275.4 cc. in seven days. One hour's reinforcement with the battery current at 62 volts developed a resistance of 2033 ohms. The figures for delivery under the same conditions as before were: 10.0 cc. in the first hour, 111.2 cc. in twenty -four hours, and 296.7 cc. in one week. When set up with a 1.3 normal sugar solution, the delivery was, in two hours, 18.9 cc., in 12 hours, 87.7 cc., and in seven days, 387.8 cc. The interest attaching to the third cell in which this membrane was deposited centres mainly in the fact that this cell, despite persistent efforts, had previously failed to show any osmotic activity when tried with the ferric hydroxide membrane. In this test, after having entirely removed the previous deposits, the cobalt ferrocyanide mem- brane proved as satisfactory as in the two previous in- stances. A resistance of 1 100 ohms was reached in one hour and thirty minutes, and the membrane showed satisfactory activity when tried with solutions of both sugar and alcohol, the velocity of delivery with a 90. 5 per cent solution of alco- hol amounting to 29 drops per minute,- 15.5 cc. in one hour, 130 cc. in twenty-four hours. This proves that the fail- ures of three cells of this type to produce a satisfactory ferric hydroxide membrane, cannot be explained by a chance selection of three cups so defective that they could not manifest activity with any membrane. It shows, too, that the firmness and consistency of some membranes may enable them to exhibit good results in walls which, with other precipitates, fail completely. We also saw earlier that the ferric hydroxide precipitate, which requires a quite perfect wall, showed remarkable activity in a cup which, with ferric phosphate, had given no hope of suc- cess. This membrane was later reinforced and set up with a thrice normal sodium chloride solution, but the delivery was quite low; only 18 cc. in twenty-four hours. The membrane seemed at the end of this time to be nearly destroyed, and the outside water, which then contained scarce more than a trace of sodium chloride, was in another 23 twenty -four hours intensely salt. When, after prolonged washing, the reinforcement of the membrane was again attempted the resistance was no higher than when begin- ning a new cell in which no membrane had ever been deposited. A fourth cell, of the same type but harder burned, showed higher resistance 1500 ohms in the first treat- ment and 3050 and 3060 ohms respectively in the first and second reinforcements. The rapidity of flow from this cell was about 20 % lower than that recorded in the case of the other cells with the same membrane, but the amounts of sugar or alcohol passing through the walls of the cell were likewise much smaller. On this subject more will be said in the general discussion later on. The Prussian Blue Membrane. This was the last membrane taken up for study in the present investigation. It was known to be an active membrane, and its capabili- ties had been somewhat investigated ; not thoroughly, how- ever, and especially not in the light of the electrolytic methods of Morse and Horn. This substance is a well- known, beautiful blue precipitate. Its ready cohesion and rather firm texture highly recommend it ; though it is of course to be expected that it will be liable, in much the same way as other ferrocyanides, to the destructive chemical action of some of the substances whose osmotic properties it may be desirable to determine. Here again a low vol- tage was found necessary, particularly at starting. In a cell of the coarser grade the attempts to secure a satisfac- tory membrane show a record which, it is thougtit, will prove not uninteresting. The voltage was varied between 4.5 and 10 volts with a N/8 solution of ferric sulphate surrounding the cell and a N/10 solution of potassium ferrocyanide within. After two hours and fifteen minutes, a resistance of 80 ohms was indicated, which remained constant during an hour of further electrolytic action. 24 Examination of the cell showed that all the precipitate had been formed within the walls. The cell when set up with a normal sugar solution showed no activity, and the follow- ing day the attempt to secure a membrane was repeated, the concentration of the outer ferric salt being made twice as great as on the former occasion. After a current of 12 volts had been passed through the cell during one hour and forty-five minutes, the resistance measured only 36 ohms. The precipitate had still made no appearance on either surface of the wall, and the feeble activity of the membrane ceased after having delivered 2 cc. in twenty -four hours. The difference in concentration of the electrolytes was still further exaggerated so that now the outer iron solution was N/2 and the potassium fer- rocyanide solution N/20. A resistance of 250 ohms was reached by four and one-half hours' electrolysis. The membrane now appeared on the inner wall, and when set up, its activity was very satisfactory indeed, a rate of about 6 cc. per hour being delivered at the beginning. This, however, after twenty hours was reduced to 0.4 cc. per hour, the delivery being of a marked blue tint. On the fourth attempt the deposition of the membrane was continued two hours and forty -five minutes. The resist- ance was 255 ohms, and the membrane, though less active, was more permanent than before, delivering 111 cc. in thirteen and a half days. A fifth time the membrane-forming process was repeated, using the same potential, 12 volts, and a resistance of 285 ohms was measured at the end of two hours' treatment. The activity and permanence of the membrane were much as before, but no blue color appeared. The sixth and seventh efforts represented about one hour each of elec- trolysis, at potentials of 62 and 105 volts, and resistances of 1080 and 1097 ohms respectively were recorded. In both cases the membrane was quite satisfactory, both in - 25 point of permanence and of activity. In the broken cell, the two membranes formed within the wall are plainly to be seen and the membrane on the inner edge of the wall is of great thickness. With the second cell, again of the coarser type, satis- factory results were more quickly obtained. A large amount of ferric hydroxide precipitate had been deposited within the walls of this cell in previous fruitless efforts to form an active membrane of that material, and without removing this, the formation of the new membrane was begun. In order to secure the thickness which seemed necessary in this membrane, a strong current was employed, as much as 0.7 amperes being passed through the cell. As in the preceding case, the results were all unsatisfactory until the relative concentrations of the electrolytes were made widely different. The concentration of the inner electrolyte was kept N/10 throughout ; the outer solution of ferric sulphate being changed from N/5 to N/2. Under the changed conditions the resistance rose satisfactorily, reach- ing 371 ohms in the first treatment, 1200 ohms in the sec- ond, and 2240 ohms in the third, when, finally, a very satisfactory membrane was obtained. Subsequent exami- nation disclosed the fact that this membrane was ex- tremely thick, local maxima reaching dimensions of 0.7 mm. With the smaller cells resistances of 17,000 and 54,000 ohms were obtained. These membranes were active with solutions of sugar, alcohol and sulphuric acid ; but in the case of the latter the. leakage was large and the delivery soon ceased. GENERAL CONCLUSIONS. The final test of all these membranes must be made in cells whose walls are capable of giving sufficient support to the membrane throughout its entire area. Until such a cell is obtained, we must conclude that the manganese fer- rocyanide membrane is unpromising both in point of per- 26 manence and activity ; that the ferric phosphate membrane is purely problematical ; that the ferric f errocyanide mem- brane is satisfactorily active, and is the most permanent of the f errocyanide membranes here tried; that the cobalt ferrocyanide has sufficient permanence for purposes of measurement, and possesses moreover very great activity, together with a firmness and consistency far superior to those of other membranes experimented with ; finally, that the ferric hydroxide membrane offers both permanence and great activity, but is incapable of manifesting the latter property except in cells whose walls are of a quite uniformly fine texture. We may now turn our attention to the consideration of such questions as whether these membranes are com- pletely or only partially semipermeable, i. e. 9 do they alto- gether prevent or merely retard the passage of dissolved substances ; whether the flow of water through the mem- brane is proportional to the difference in concentration of the solutions separated by the membrane; whether the diminution of concentration within the cell bears any fixed ratio to the volume which it has delivered and whether the deficiencies observed are due to defective cell walls; and, if so, can they be remedied. In answering the first question it must be said that in every one of the many cases observed there has been leakage through the walls of the cell. Hence, either the membrane or the wall is defective, perhaps both. The evidence all points towards the deficiency of the wall. This evidence is, first, the fact that in all cases in which the deposition of the membrane was long continued a maximum resistance was reached, which ordinarily became constant for all further treatment. This fact seems capable of explanation only on the supposition that there is a constant cross-section of unimpeded transference of the ions ; or, in other words, a number of pores too large to be bridged over by the membrane. 27 Second. The much larger leakage when the cell is deliver- ing under somewhat greater pressures. Three examples taken from a number of these observed differences will suffice. Cell XXII. was tried with normal sugar solutions, while in cells XVII. and XV. exactly 60.0 gms. of sugar was employed in each instance. The periods of time, pressures in millimeters of water, and leakages are given below. CELL XXII. Time. Pressure. Leakage. 7 da , 50 min. 10mm. 5.074 gms. 7 " 70 " 74 " 7.052 CELL XVII. Time. Pressure. Leakage. 10 da., 21.5 hrs. 10 mm. 4.532 gms. 10 " 20 " 174 " 6.587 " CELL XV. Time. Pressure. Leakage. 10 da , 22 hrs. 10 mm, 4.354 gms. 10 " 19 " 222 " 7.077 " Now, if the membrane is intact this leakage is simply the measure of the diffusion of the dissolved substance ; and the rate of diffusion of a given substance through a given medium, the temperature remaining unchanged, de- pends only upon its concentration. But the concentra- tion is not appreciably altered by these slight differences of pressure. Consequently, not diffusion merely, but a positive outward flow of the solution takes place. Hence there are interstices which make this movement possible. Third. The same membrane, cobalt ferrocyanide, pre- pared with equal care, gives in different cells widely dif- ferent amounts of leakage. The thickness of the semi- permeable membrane should, if the area is the same, vary with the number of coulombs represented by its formation. Hence, in cell XXI., in the formation of whose membrane 0.19 ampere hours were used, the membrane should be not appreciably thicker thstn in cell XIX., whose mem- brane represents 0.17 ampere hours of current. And if 28 the leakage is to be explained by diffusion there should be no wide difference here. Yet, the leakage through cell XXI., viz., 0.8904 grms. sugar in 7 days, 23 hours, was much less than through the other wall. The number for this is 5.458 grms. sugar in 12 days, 2^ hours. Again, for cell XXI. set up with a solution containing 74.33 grms. alcohol we have the small leakage of 0.2077 grms. in 11 days, 23 hours; while in cell XX. with the same mem- brane and set up with 127.2 grms. of alcohol we have a leakage of 3.540 grms. in 10 days, 18 hours. The same cell, XXI. with the advantage, however, of a reinforce- ment of the membrane shows a smaller leakage with the much smaller and more active alcohol molecule than that recorded for the sugar ; while if diffusion were the explana- tion, the opposite result should have been observed. The figures for cell XXI., both in point of delivery and smallness of leakage, show that this membrane is semi- permeable quite as well for the smaller alcohol molecule as for the larger molecule of sugar. The evidence, such as it is, favors the complete semipermeability of the membrane; though a demonstration is hardly possible with the means at hand. The answer to the question: Is the flow through the membrane, of the pure solvent from without, proportional to the concentration of the solution within? seems to be the same both in theory and practice. The rate of flow in- creases with higher concentrations, but not in a direct ratio. This is to be expected if we consider that the porous wall offers a very high resistance to the rapid pas- sage of a liquid. Hence, the greater the concentration and the more rapid the flow, the more effectively will this resistance be felt. Again, the rapid influx of water through the membrane dilutes more appreciably the solu- tion immediately adjacent to the membrane, thus reducing the effective concentration of the solution, and, in con- sequence, the flow through the membrane. Further, there is a dependence upon the relative densities of the solu- tions employed. In the sugar solution the density exceeds that of the outer water, and the dilution of the portion adjacent to the membrane makes this part lighter than the remaining liquid, giving, in consequence, upward currents along the cell wall toward the point of delivery. The solution delivered is, therefore, of lower density than the average concentration remaining within the cell. While if alcohol be used, the opposite results must take place, the more concentrated solution being first delivered, and the greater density of the lower concentration adjacent to the membrane causing this to flow downward. All these will have their influence in producing a rate of delivery somewhat different from that which should result from a uniform concentration of the solution within the cell. If now we turn from theory to the figures observed, we find the flow from cell XVI. at a concentration N/45 to amount to 7.7 cc. per day. The proportional flow for a normal solution would be more than 346 cc. per day under the same conditions of temperature. These temperature conditions were iden- tical when seven days later with a fresh solution of normal sugar this same cell exhibited an initial flow of 293 cc. per day a difference quite appreciable, yet, perhaps, not so great as we should expect with these large figures. The measurements from which comparisons are drawn, must, unless corrections can be applied, be taken when the outer solvent is pure. For instance, in cell XVI. at a concen- tration N/45 the flow was twice as great as when this concentration was N/61, because of the presence of a very small amount of sugar in the outer water at the latter concentration. This same cell at a concentration 0.03 grms. per cc., when the amount of sugar in the outer water had risen to 0.020 grms. per cc., trebled its flow on 30 replacing the latter by pure distilled water. The differ- ence of the concentrations within and without had been 0.01 grms. per cc. in the former instance, and was three times that after the change. This showed that at low concentrations the flow is very nearly in direct proportion to the difference of concentration of the inner and outer liquids; also that unless the outer water is pure, or its concentration known, the flow from a given cell cannot be taken as a measure of the activity of its membrane at the actual concentration of the solution within ; the more so if this concentration be rather low, for then the concentra- tion of the outer liquid may easily represent a large per- centage of the concentration within, and the flow through the membrane will be diminished in the same ratio. As the leakage from the cells was in most cases allowed to accumulate and no accurate knowledge of its concentration was attempted until several days had passed, the many readings taken represent accurately the relative activities of the various membranes at the beginning only of the record and at the end when it was corrected for the outer concentration then measured. Still all the earlier readings taken while the concentration of the outer liquid was an inappreciable fraction of the high concentration within, are close approximations to the correct values. Hence, the curves of several of the typical cells are here given. The total delivery is plotted against time and the slope of the curve represents the rate of flow of the cell for the corresponding time. Curve I. illustrates the action of a cobalt ferrocyanide membrane in cell XIX. when set up in a constant tempera- ture bath at 35. The sugar solution was 1.3 normal. Curve II. is for the same cell at the same temperature, but with a solution exactly normal. Curve III. represents the activity of the same membrane in Cell XVI. when set up with a normal sugar solution at the temperature of the laboratory about 18. 31 Curves IV. and V. are for the Prussian Blue and Man- ganese ferrocyanide membranes. The conditions are the same as those last given. These three are good types of the records made by the respective membranes. 32 CONCENTRATION CHANGES. The relation between the total volume delivered and the diminishing concentration within the cell is also interest- ing. If we suppose diffusion to prevent differences of con- centration in the different portions of the cell, this relation is easily worked out as follows : Let V = the capacity of the cell and v the total volume delivered. Then v is the independent variable and the concentration, K 9 is a func- tion of v and may be written K '= (v). The total dis- solved substance in the cell at any time is, then, V 4>(v) the amount at starting being equal to V (o). Now when v receives the increment dv, the substance lost by the cell is (j>(v)dv. This quantity can also be expressed by representing the difference between the total amounts of dissolved material in the cell before and after the delivery of the volume dv. Hence we have = ~ Whence log e (v) = + (7. Now making v = 0, we have log e (0) = C y , and subtracting, log e ( loge T t vof rimed on time are subject to a fine of Books not ret "5 ne< * xTTi-.j /io V overdue increasing 50c per volume ^^^Ir the sixth day Books not i: expiration of loan period.