]\gt ; Thae Selective Properties of the Copper Membrane .
* By FRANK TINKER , D.Sc .
Communicated Oliver Lodge , Received In a previous researoh dealing with the structure of semi-permeable membranes , it has been shown that the copper fe1rocyanide membrane is built up from colloidal particles diameters ranging from 50 to ; also that the pores enclosed by these particles have an aveIage diameter which , although large to allow , to any non-colloidal molecule , is small to be entirely within the range of capillary effects .
On the basis of these facts the hypothesis was advanced that the selective properties of colloidal membralles are due to selective adsorption ; a copper ferrocyanide membrane immersed in cane-sugar solution , for instance , being permeable to the water and not to the cane sugar , because the water , rather than the sugar , is selectively adsorbed or " " condensed\ldquo ; on to the membrane surface and into its capillaries .
The present research is an experimental ation of this view .
The measurements made hereill consist in a determination of the change in concentration about by immersing more or less dry colloidal copper ferrocyanide in cane-sugar solutions of various .
The results obtained tend to confirm the adsorption hypothesis for this case at least .
I. Experimental Method .
The method of experiment usually adopted was to shake up 10 .
of colloidal copper ferrocyanide prepared as below with 100 .
of solution for two days at room temperature At the end of that time , the solution was cleared completely of the finely divided ferrocyanide with the aid of a powerful Delepine centrifuge developing a force of about 10,000 times gravity .
Its optical rotation was then determined side by side with that of a blank sample of the original solution which had been treated similarly to it at all , except for the introduction of the ferrocyanide .
The colloidal copper ferrocyanide was prepared as follows : solution of 215 .
of potassium ferrocyanide in litres of water was added to a solution of 250 .
of copper sulphate , also dissolved in litres of water .
The mixturc was then diluted to 5 litres , churned up for a few hours , and * Being part of a Thesis approved for the Degree of Doctor of Science in the University of London .
Tinker , 'Rov .
Soc. Proc , vol. 92 , p. 357 ( 1916 ) .
Slight changes in temperature of one or two degrees centigrade have no measurable influence on the adsorption equilibrium .
Selective Properties of the Copper Ferrocyanide .
269 the copper ferrocyanide suspended in it then centrifuged down .
In order to free the colloid from most of the salts ( mainly ) entangled in i was washed twice by churning with 5 litres of water for one day , the ferrocyanide separated each time by the as before .
At this the colloidal ferrocyanide was in the form of a sticky clay .
It was converted into a dry clay by it up in filter-paper pockets placed above absorbent lint in a series of centrifuge tubes ( fig. 1 ) , and then centrifuging for a few hours .
It was dried still further by steeping it in anhydrous ether for two days , the ether left attached to the ferrocyanide after decantation being removed by current of dry air for one day over the colloid , kept at a temperature of from C. to C. So prepared , the ferrocyanide was obtained as a brown , fluffy powder , which , although not absolutely dry , was found to take np moisture if left exposed to the atmosphere .
yiouS t use in the subsequent adsorption experiments , it was dried more thoroughly in the manner indicated in each case .
Inasmuch as the changes in the con centration of ) bulk of a solution due to adsorption are not very large , it was found necessary to introduce a of elinement iuto the polarimetric measurements .
dingly , llany of the improvements introduced into work by Dr. T. M. were adopted .
The source of ligl ) , for instance , was a 600-c.p .
Cowper-Hewitt silica mercury vapour lamp , ab modified by Dr. Lowry .
The beam from it was condensed without previous resolution into a polarimeter , and separated into its components after passing through the sugar solution by means of a small direct vision spectroscope placed in the polarimeter iece .
The for ths mercury reen line only were taken , the colunm of solution being 6 .
in each case .
The temperature at which the rotations were determined was ) C. , the Lowry bath , and system being employed .
S It was found that the * It is fatal to wash the colloid too thoroughly .
If all the entangled salts are completely removed , the ferrocyanide subsequently forms with the sugar solution a sol which the centrifuge brings down only with extl.eme difficulty .
The details of the preparation given above are all more or less essential for success .
Unless care is exercised the ferrocyanide is apt to be obtained on drying in the form of a horny mass which is tough , highly compact , unwolkable .
Lowry , ' Trans. Chem. Soc vol. 103 , p. 1063 ( 1913 ) 'Proc .
Roy .
Inst Apri118 , 1913 , and other papers .
S For a working knowledge of these things I am indebted to the valuable training Dr. Lowry gave me as a student with him .
I am also much indebted to Sir Olivel Lodge for the provision of the instruments and materials in question .
Dr. F. Tinker .
The Selective solutions could be read quite easily to , so that the experimental error on a difference of one or two degrees rotation was within one per cent. In order to prevent in the concentration of the solutions due to fermentation , all vessels were invariably sterilised , and the solutions themselves rendered sterile before using by raising them to C. for a few minutes .
Preliminary to the concentration experiments proper being carried out , the following experimental points were also ained : ( a ) When the colloidal fel.rocyanide is placed in the solution , equilibrium is attained within a hours .
A series of 10-grm .
samples of ferrocyanide placed in 100 .
of 20-per-cent .
sugar solution ( rotn .
in a 6-dm .
tube at ) gave rotations of after 5 hours , 1 day , 2 days , and 4 days respectively .
( b ) Different samples of ferrocyanide from the same stock alter the rotation of any solution to almost but not quite the same extent .
Thus , four 10-grm .
samples from the same bulk increased the rotation of 100 .
of 20-per-cent .
sugar solutions from ( in a 6-dm .
tube ) to respectively .
This difference is probably due to inavoidable inequalities in the colloidal condition of the ferrocyanide , even in similar samples .
Very regular curves , connecting the change in concentration with the concentration itself , are therefore not to be expected .
( c ) Neither the colloidal ferrocyanide itself , nor the traces of potassium sulphate left in it , exert any hydrolysing effect on the sugar solutions under the conditions employed .
II .
rimental Results .
In obtaining the following results , different bulk samples of ferrocyanide were used in the different series of experiments .
It will be seen that these samples differed somewhat from one another in degree of dryness , according to the method of drying adopted .
The object of varying this factor was determine how the changes in colloidal condition induced thereby would affect the results .
The Tables are built up on the assumption that water only and no sugar is absorbed by the colloidal ferrocyanide .
* On this assumption , it is evident * This assumption is not quite correct .
Sugar will be " " adsorbed\ldquo ; to a certain extent also , at least into the outer layers of the surface ] .
In this case , if is the weight of sugar absorbed from 100 , and is the concentration of the solution ( grammes per 100 ) is thus really greater than the Tables would indicate .
Properties of the Copper Membrane .
that , if .
is the weight of water removed from 100 .
of the solution by shaking with the ferrocyanide , Rotation at equilibrium rotation , whence .
The last column in the Tables , headed " " water absorbed by 100 dry ferrocyanide is calculated from the previons column by allowing for the amount of moisture inally in the colloid before immersion in the solutions .
Thus , in Table II ( bulk sample B ) , the true veight of the 10 of ferrocyanide taken in each case is , and in this case the column is given by the relation .
Total water absorbed at equilibriunl Table I.\mdash ; Bulk Sample A of Copper anide .
Dried one day in steam oyen .
10 grm. contained .
water absolutely ) .
October 7 , 1915 .
4060 *The strength is given in grammes cane ugar added to 100 ?
uater .
The rotations are those given by a 6 .
tube .
Table II.\mdash ; Bulk Sample of Copper errocyanide .
Partially dried over conc . .
contained .
water .
October 12 , Dr. F. Tinker .
The Selective Table III.\mdash ; Bulk Sample of Copper Ferrocyanide .
Partially dried by rrent of air at C. for three hours .
10 .
contained water .
November 29 , 1915 .
Table \mdash ; Bulk Sample of Copper Ferrocyanide .
This sample was not dried at all .
10 .
contained .
water .
It will be noted that the solutions were weakened in this case .
In spite of this , however , the total water absorbed by 100 .
of ferrocyanide is of the same order as in the other three cases .
September 7 , 1915 .
Approximatestrength oOrigi.iumBotatio.ovedfro .
shows graphically the relation between the concentration of the solution and the total water orbed by 100 .
of dry ferrocyanide for each of the four samples chosen .
It will be seen that water is absorbed by the ferrocyanide from all the sugar solutions to quite a considerable The amount absorbed decreases as the strength of the solution increases , and varies from about 30 .
of water from pure water to 13 or 14 .
of water from a 60-per-cent .
solution of cane sugar .
* The curves show no tendency , however , to approach the axis , but rather to become horizontal Sample a somewhat higher curve than the other three samples .
Further experiments with fresh samples that this behaviour is characteristic of samples dried slowly in air high temperatures The samples dried like this turn almost black , apparently because some of the ferrocyanide is decomposed into copper oxide .
The results for this sample therefore not so reliable as for the other three samples .
Properlies of the Copper Ferrocyanide Membrane .
when the moisture-content is about .
in the case of strong solutions .
This fact would indicate that the ferrocyanide forms a hydrate STRENGTi{ SUGA SOLUTION .
FIG. 2 .
which corresponds to a moisture-content of .
water per 100 .
dry ferrocyanide .
The probability that copper ferrocyanide forms a stable hydrate in all except the very aqueous solutions was noted by Graham on rounds quite different from the present one .
* the existence of this stable bydrate in solutions of all stren yths , we have now to explain in what manner the moisture in excess the amount which corresponds to this hydrate is taken , whether by extra hydration of the copper ferrocyanide molecules , or by solution in the colloidal phase , or by " " adsorption\ldquo ; proper on to the of the colloidal particles .
It will be seen that the facts are in agreement with the adsorption hypothesis ; for there is no simple relation between the amount taken up and the conccntration ; neither is the amount taken up by a given weight of dry ferrocyanide from a solution of given strength a constant , as it should be if the water were either dissolved water or hydrate water .
This latter variation in amount of water taken up can only be due to variation in the surface area of colIoid in the different samples , causing a corresponding increase or decrease in the adsorbing capacity .
III .
Application of the to the Theory of The above experiments bring out several facts with regard to the copper ferrocyamde membrane .
*Graham , " " Liquid Diffusion applied to Analysis ' Collected Papers , ' p. 583 .
Dr. F. Tin The Selective ( 1 ) The membrane is really composed of colloidal particles of copper ferrocyanide hydrate , probably ( 2 ) The colloidal particles of copper ferrocyanide hydrate in the membrane adsorb water selectively from solutions of cane , without taking up the to auy appreciable extent at the same time .
( 3 ) The amount of water adsorbed depends on the strength of the sugar solution in which the membrane is placed , decreasing as the solution gets stronger .
The selectiye adsorption hypothesis previously advanced as the reason for selective permeability follows immediately from the second of these facts .
It is self-evident that if the sugar cannot get into the membrane because of the selective surface effects , neither can it get through the membrane for the same reason .
Likewise from the third fact we can arrive at the cause for osmotic flow .
If pure water , for instance , is on one side of a copper ferrocyanide membrane and solution is on the other , the adsorbed moisture-content of the membrane and , therefore , the adsorbed moisture pressure also , will be reater on the water side of the membrane than it is on the solution side .
A continual across the membrane from the water to the solution will therefore take place , equilibrium being impossible so long as there is a concentration difference between the liquids bathing the two sides .
* It may be mentioned in passing that , in principle , this view of the cause osmotic flow was first advanced by Graham , who showed that gelatinous membranes swell more in contact with pure water than they do in contact with certain salt solutions .
Graham explained osmotic flow by the hypothesis that the water side of the membrane is more highly " " hydrated\ldquo ; than the solution side .
In this connection it is also interesting to note that similar results to the above have also been obtained by Prof. Adrian Brown with the seeds of common barley .
These seeds form ideal osmotic system , in that they are really a mass of colloidal material starch ) , surrounded by a semi-permeable ulembrane .
Prof. Brown has shown that , when the seeds are immersed in aqueous solution of cane sugar , common salt , and other solutes impermeable to the membrane , the amount of water absorbed by the colloidal contents is less than that absorbed by the seeds from pure water , being diminished to about one-half *I.e .
, in the absence also of a pressure being applied to the solution in order to prevent diffusion .
Phil. Trans 1861 , pp. 183-224 Liquid Diffusion applied to Analysis Section 8 ) .
This view has also been revived by Trouton ( Presidential Address , British Association , Section , 1914 ) .
'Roy .
Soc. Proc , vol. 81 , p. 82 ( 1909 ) .
Properties of the Copper Ferrocyanide the latter amount in the case of molar solutions .
Prof. Brown and the present author have , however , lately shown* that the amount of solvent taken up by the seeds depends also on the amount of solute taken up simultaneously , being at a minimum when the membrane is truly semipermeable to the solute , equal to the amount taken up from pure water when the membrane exerts no selective action , and actually than the amount taken up from pure water when the seeds adsorb the solute selecGiveIy .
In the of the ) facts , it is evident that we can no longer regard the actual colloidal membrane as a geometrical plane perforated with small holes .
It is rather a separate phase in the osmotic systen ] , into which the solvent can diffuse from both the pure solvent and solution .
The solvent flows from the pure solvent to the solution , because the former , enerates a greater solvent pressure and concentration inside the membrane than does the latter , just as it also enerates a greater concentration and pressure inside the vapour phase .
The phenomenon of osmotic flow thus becomes similar in nature to that of vapour flow , and both have their origin in the greater " " activity\ldquo ; of the pure solvent .
But , although we are justified in concluding that the two phenomena are similar , we must not assume that they are identical .
In ] , we must guard against that the solvent travels across a semi-permeable membrane as vapour , for the experimental work which has been described shows that the concentration of the solvent inside the membrane is much greater than it is inside the vapour phase proper , to the iact that the colloidal particles of the membrane condense it on to their surfaces .
The solvent travels across the membrane in the adsorbed condition , and as such is neither liquid nor vapour .
IV .
The selectiye properties of copper ferrocyanide have been studied by measuring the change in solution concentration which takes place when the dry colloid is immersed in cane-sugar solutions of various .
It is found that the solutions become stronger , owing to the fact that the water and not the sugar is taken up selectivel .
) the fel.rocyanide .
The experimental results are in accordance with the hypothesis that a colloidal hydrate is first formed , and that this colloidal hydrate takes up still more moisture by adsorption .
The amount of ' adsorbed\ldquo ; moisture taken up by the colloid decreases as the strength of the solution increases .
The fact that colloidal ferrocyanide adsorbs water selectively from cane'Roy .
Soc. Proc , p. 373 ( 1916 ) ; ibid. , , vol. 89 , p. 119 1915 ) .
Cf .
Callendar 's vapour pressure theory .
Prof. W. H. Young .
On the Ordinary solutions without taking up the sugar at the same time supports the author 's theory that the property of sele tive permeability is ( in the case of colloidal membranes ) a result of preferential adsorption .
It is also shown in the above paper that the side of a membrane in contact with pure water has a greater moisture content than the side in contact with sugar solution .
This fact supports the hypothesis\mdash ; first .
advanced by Graham on experimental grounds\mdash ; that osmosis across the lnembranc takes place because pure water induces a greater moisture .
ressure and concentration inside the membrane than the solution does .
On the Ordinary rgence of Restricted Fourier Series .
By Prof. W. H. YOUNG , Sc. D. , ( Received October 11 , 1916 .
) S1 .
The necessary and sufficient condition that a trigonometrical series* , Qhould be a Fourier series is that the integrated series should converge to an throughout the closed interval of perioaicity , and should be the Fourier series , , of an integral .
Conversely , starting with Fouriel series of an integral and differentiating it term by term , we obtain the ouriet series of the most general type , namely , one associated with any fdnCtion possessing an absolutely convergent integl'aL If the Fourier series which is differentiated is not the Fourier series of an integral , but of a function which fails to be an , at eve1l a single point , the derived series will not be a Fouriel series .
series so obtained , namely , by term-by-term the Fourier series of a function which , without being an integral the whole closed interval of periodicity , is an integral in one or more sub-intervals , I have founJ it conyenient in a recent to call ' restricted Fourier series I retain the ' Fourier because such series possess , in the interval or * It is convenient to suppose the trigonometrical series to have no constant term , or to egard the integrated serie as a series obtained from the original series by interchanging coefficients of and , changing the sign of one of them , and dividing both by .
The integral to which the integrated series converges is accordingly a periodic function .
By integral we mean , as usual , absolutely convergent or Lebesgue integral .
" " On the Convergence of the rived Series of Fourier Series read at the June Meeting , 1916 , of the L.M. S. and recently presented to that Society .

