557 An Experimental Investigation of Gibbs ' Thermodynamical Theory of Interfacial Concentration the Case of an Air- Water Interface .
By Prof. F. G. Donnan , F.R.S. , and J. T. Barker , D.Sc .
( Received July 31 , \#151 ; Read November 9 , 1911 .
) ( From the Muspratt Laboratory of Physical and Electro-Chemistry , University of Liverpool .
) 1 .
Introduction.\#151 ; Consider an aqueous solution of a substance S , the concentration being denoted by c. Let this solution be separated from another phase ( oil , air , etc. ) , in which the concentration of the substance S is negligibly small .
We may apply to this case Gibbs ' fundamental equation* p __ do " ~ d~f where a is interfacial tension , ^ is chemical potential of S in the aqueous solution , r is mass of S per unit area of interface in excess of that corresponding to the uniform bulk-concentration of S in the solution .
It will be seen that T is , in fact , the amount of S per unit area of interface adsorbed or concentrated in the interfacial transition layer .
If we are justified in applying the laws of dilute solutions , the above equation can be written .t " , C r = ~ rt where R is the constant of the simple gas equation and T is absolute thermodynamic temperature .
This equation is of fundamental importance in the study of the phenomena of the surface-accumulation of substances at the surfaces of separation of different phases .
We may regard it as giving us a means of calculating the quantity T , the " capillary adsorption " at the interface .
If da/ dc be negative , r will have a positive value , whilst if dafdc be positive , T will be negative , that is to say , the concentration of S at the interface will be less than in the bulk of the solution .
In an experimental test of the validity of the equation , we must clearly deal with a freely mobile interface , in order that both T and da/ dc may bn measured .
* Compare ' Scientific Papers , ' vol. 1 , p. 219 , et seq. VOL. LXXXV.\#151 ; A. 2 Q 558 Prof. Donnan and Dr. Barker .
Gibbs Thermo- [ July 31 , At the suggestion and under the direction of one of us , W. C. McC .
Lewis* has investigated in this laboratory the case of a liquid-liquid interface , one liquid being an aqueous solution , the other a hydrocarbon oil .
In the case of substances such as sodium glycocholate , sodium oleate , Congo red , and methyl orange , the directly measured value of V was found on an average to be about 60 times greater than the calculated value .
It will be noticed that these are substances of high molecular weight , and of marked colloidal character in ( concentrated ) aqueous solution .
The discrepancy is therefore most probably due to the occurrence of complex changes in the concentrated surface-layers , whereby colloidal aggregates , and possibly gelatinous or semi-gelatinous " skins , " are formed .
Smaller discrepancies were observed in the case of some pimple inorganic salts , and in the case of caffein there was practical agreement between the observed and calculated values of T. Unfortunately , however , the adsorption was so small in these latter cases that the results partake more of a qualitative than a quantitative character .
Recently , f the adsorption of aniline at the interface between an aqueous-alcoholic solvent and mercury has been measured by Lewis in a similar manner , and found to be in good agreement with the value calculated from the rate of variation of the interfacial tension with concentration .
In these experiments , however , the effects measured were still not very great in comparison with the possible experimental error .
It was decided therefore to investigate the phenomenon of capillary adsorption in a case where the effects would be as great and the conditions theoretically as simple as possible .
2 .
Choice of Substances.\#151 ; On account of its high surface-tension , water was chosen as the solvent .
In order to make the conditions , theoretically at least , as simple as possible , it was decided to investigate the adsorption ( or surface-concentration ) at the interface between water and air .
It was necessary , therefore , to select a substance which should satisfy the following conditions:\#151 ; ( 1 ) It must be of simple and definite chemical composition , and give true aqueous solutions of simple and definite character .
( 2 ) It must be , under the conditions of the experiments , practically nonvolatile , and sufficiently soluble in water .
( 3 ) It must in extremely dilute solution exert as powerful an effect as possible in lowering the tension of the air-water interface .
A search of the existing data showed that one of the saturated aliphatic carboxylic acids would be likely to satisfy all these conditions , and our choice finally fell on nonylic acid ( pelargonic acid , CgHigOs ) .
The lower * ' Phil. Mag. , ' April , 1908 , p. 499 , and April , 1909 , p. 466 .
t 'Zeitschr .
fur physikal .
Chemie/ 1910 , vol. 73 , p. 129 .
1911 .
] dynamical Theory of Interfacial .
559 rn ) c homologues of this series are too volatile , the higher not sufficiently soluble in water .
We chose nonylic acid as sufficiently satisfying both the conditions specified in ( 2 ) .
The liquid acid was obtained from Kahlbaum , and its purity examined and tested with satisfactory results .
Some preliminary measurements were made with solutions of sodium taurocholate , sodium glycocholate , and the corresponding free acids .
The latter were found to be much more active than the salts .
But since nonylic acid was found to be markedly more active than any of these substances , and possessed the further advantage of greater chemical simplicity , it was chosen as the best substance for our purpose .
3 .
Previous Investigations.\#151 ; The theory of surface-concentration at an air-water interface has been discussed by Lord Rayleigh , * especially in connection with the stability of foams .
Lord Rayleigh 's experiments on the surface-tension of a newly-formed aqueous soap solution and air interface have shown that an appreciable time is required , for the formation of the more concentrated surface-layer .
The existence of this surface-concentration in the case of an air and water interface has been demonstrated by the experiments of D. H. Hall , " !
* J. von Zawidski , J F. B. Kenrick and Miss C. Benson , S and S. R. Milner.ll No satisfactory quantitative investigation of the phenomenon has , however , been carried out as yet .
4 .
Analysis of Solutions and Determination of \#151 ; For this purpose a dropping-pipette of the form shown in fig. 1 was employed .
The simple theory of the instruments is that for a given size of opening the surface-tension is directly proportional to the weight of a drop , and hence , for very dilute solutions of practically equal density , inversely * Cf , for example , * Proc. Roy .
Inst. , ' 1890\#151 ; 1892 , vol. 13 , p. 85 .
+ ' Proc. Roy .
Dublin Soc. , ' vol. 9 ( N.S. ) , Part I , No. 10 , p. 56 .
X ' Zeitschr .
fur physikal .
Chemie , ' 1900 , vol. 35 , p. 77 .
S ' Journ. Phys. Chem , , ' 1903 , vol. 7 , p. 532 .
|| ' Phil. Mag. ' [ 6 ] , 1907 , vol. 13 , p. 96 .
IT Lord Rayleigh , 'Phil .
Mag. ' [ 5 ] , 1899 , vol. 48 , p. 321 .
Fig. 1 .
560 Prof. Donnan and Dr. Barker .
Gibbs ' Thermo- [ July 315 proportional to the number of drops in a given volume .
This holds strictly only for static equilibrium , the dynamic nature of the actual phenomenon necessitating a small correction.* This correction is unnecessary when we employ the drop number-concentration curve as a means of analysis .
Even in the case of the determination of from the curve we have employed only the simple ( statical ) theory , as we considered that the correction would not be large enough to sensibly affect our main conclusions.* !
* It is necessary , however , that the rate of formation of the drops should be sufficiently small , as pointed out by Lord Kayleigh .
In the case of dilute solutions of nonylic acid , where there is strong surface-concentration , the surface-tension of the freshly formed surface takes an appreciable time to arrive at its equilibrium value , and so the number of drops in a constant volume ( the volume of the pipette ) varies considerably with the speed of dropping .
In order to obviate this source of error , which , if neglected , would have rendered our analytical results quite worthless , the speed of dropping was kept quite constant ( or very nearly so ) in all measurements ( 8*3 drops per minute ) .
Under these conditions we found we could reproduce the drop-number of a solution of nonylic acid of given concentration to a fraction of a drop ( in 300 to 500 drops ) .
This procedure rendered the use of the dropping-pipette quite satisfactory for the purposes of analysis .
For the determination of relative surface-tensions it would , strictly speaking , have been necessary in the case of every solution to have determined the drop-number as a function of the speed of dropping , and to have extrapolated to zero speed .
The ( a , c ) curve determined by us for the constant speed of 8'3 drops per minute will not therefore quite coincide with the ( a , c ) curve determined for an infinitely slow rate of dropping .
The small error thus introduced in the values of may be neglected in the present investigation , where we are concerned more with orders of magnitude than with high percentage accuracy .
The pipette shown in fig. 1 was water-jacketed to keep the contents at a constant temperature .
The forming drop must be carefully shielded from draughts and vibration .
The orifice consisted of a thin-walled cylindrical tube of about 2 mm. diameter sealed to the main stem .
The external walls of the orifice must he kept dry and clean .
The solution to be measured is sucked up into the pipette by means of the side tube , the tap of which is afterwards closed .
The speed of outflow is regulated by fine capillary tubing * Guye and Perrot , 'Archives des Sc. Phys. ' ( 4th series ) , 1901 , vol. 11* , p. 225 ; 1903 vol. 15 , p. 132 .
Kohlrausch , ' Ann. der Physik .
, ' vol. 20 , p. 798 ; 1906 , vol. 22 , p. 191 .
Lohnstein , ' Ann. der Physik .
, ' vol. 20 , pp. 237 , 606 ; 1906 , vol. 21 , p. 1030 .
t Cf .
Lewis , ' Phil. Mag. , ' April , 1908 .
1911 .
] dynamical Theory of Interfacial Concentration .
561 attached to the top .
Entrance of dust is prevented by suitable plugs of cotton wool .
The number of drops was automatically registered on the tape of a Morse recorder by allowing each drop to fall on a balanced vane of aluminium , the movement of which closed the relay circuit .
Towards the end of the dropping the drops were counted by eye , in order to estimate to a fraction of a drop .
The reliability of the automatic registration was frequently checked by direct counting .
The solutions of nonylic acid were made up by weight in distilled water , and were preserved in vessels of Jena glass which had been previously carefully cleaned and steamed out .
Since sodium nonylate , like the sodium salts of the bile-acids , is much less active in lowering the tension of the air and water interface than the free acid , the drop-number of the solutions constitutes a very delicate test for their permanence in contact with glass .
It was found that the weaker solutions remained unchanged for several days , whilst the stronger solutions showed a gradual fall in drop-number .
Fresh solutions were therefore always used in these cases .
The following table gives the measurements of drop-number and surface-tension , as determined by us for solutions of nonylic acid ( lfi'S0 C.):\#151 ; Table I. Concentration ( c ) .
Drop-number .
Surface tension ( \lt ; r ) .
grm. in 100 grin , solution .
dynes/ cm .
0 324-8 72 -91 0 -000977 329 -5 71 -83 0 -001260 330*7 71 -65 0 -002430 348 -6 67 -97 0 -003290 364 -5 65-01 0 -005007 " 1 414-01 0 -005000 y 413 0 l 57 -33 0 -005005 J 413-Oj 0 -006810 449-0 52 -77 0 -007590 471-7 60 -24 0 -008060 482-7 49-09 The corresponding ( a- , c ) curve is shown in fig. 2 .
The most remarkable feature of this curve is the point of inflexion , which occurs at a concentration of about 0-004 per cent. This has also been observed by Forch , * who has determined the ( \lt ; r , c ) curve for nonylic acid and its homologues .
It might be possible to ascribe this effect to the presence of small amounts of ammonia in the distilled water , since the ammonium salt would not be so active as the free acid .
It seems more probable that the above form of * ' Wied .
Annalen , ' 1899 , vol. 68 , p. 801 .
562 Prof. Donnan and Dr. Barker .
Gibbs ' Thermo- [ July 31 , the ( \lt ; r , c ) curve really represents the true relation between a and We have , however , directly measured V for several values of c lying between 0*0024 and 0*0080 per cent. , in order to eliminate any doubt as to the true values of derjdc corresponding to very small concentrations .
For O IO 20 30 40 50 60 70 80 90 XOO Concentration(in ten-thousandths of I percent ) of Nonylic Acid ( Temp.=i6-5 ' ) Fig. 2 .
values of c higher than 0*0024 per cent. , it seems probable from both Forch 's curve and ours that the corresponding values of would not be subject to the possible sources of error which might be effective in extremely dilute solutions , especially of acids .
5 .
Experimental Determination of F.\#151 ; The principle of the method employed is as follows :\#151 ; A large number of air-bubbles of uniform and known size are allowed to traverse a column of the solution .
These bubbles carry on their surface a layer of more concentrated solution , due to the effect of the capillary adsorption at the air and solution interface .
The bubbles are caused to collect ( and break ) in another vessel , so that the excess of solute thus returned to the bulk of the solution cannot find its way back to the column of solution referred to above .
If we know the number and size of the bubbles and hence the total interface produced , and can determine with sufficient accuracy the total amount of solute removed as a result of the 1911 .
] dynamical Theory of Interfacial Concentration .
563 excess surface-concentration on each bubble , we shall clearly be able to calculate Ydirectly .
Since we must work with very dilute solutions , in order to make T/ c as large as possible and'since the change of concentration produced by thousands of air-bubbles is , even in the case of very active substances , very small , an essential condition of success is the possession of a very delicate means of estimating these small differences of concentration of very dilute solutions .
Ordinary chemical methods of analysis would quite fail us here .
Fortunately the dropping-pipette itself provides us with the necessary weapon , owing to the rapid change of drop-number with concentration in the very region where the change of concentration has to be determined .
It is only necessary therefore to determine the drop-number for the given solution before and after the bubbling , in order to be in a position to estimate the small decrease of concentration produced .
A reference to Table I will serve to illustrate the degree of accuracy obtainable .
For a 0*005 per cent , solution of nonylic acid a decrease in the drop-number of 1 drop corresponds approximately to a diminution of concentration of 0*000034 grin , per 100 grm. of solution .
As we can estimate to half a drop with considerable certainty , it is possible to detect a change in concentration of 0*34 per cent , of its value in a solution containing only 0*005 grm. nonylic acid in 100 grm. This degree of sensitiveness of the " capillary " method of analysis has made the measurements described in this paper possible .
The average drop difference observed amounted to about 4 drops , corresponding to a change of about 2*7 per cent , in the solution containing 0*005 grin , nonylic acid in 100 grm. Putting the mean error of measurement at about one quarter of a drop , the above change of concentration could be estimated with an accuracy of about 6 per cent. The chief difficulty encountered in the determination of T arose from the streaming of the liquid caused by the sweeping action of the rapidly moving bubbles .
This streaming had the effect of mixing the liquid in all parts of the apparatus , and thus preventing the localisation of the decrease of concentration which we desired to measure .
It was only after trying many forms of apparatus without success that we hit upon the simple device of breaking up the steady streaming motion of the liquid into a series of localised eddies , by means of a sort of inverted fractionating column .
The form of apparatus finally employed and the method of using it will be made clear by a reference to fig. 3 .
564 Prof. Donnan and Dr. Barker .
Gibbs Thermo- [ July 31 , Air at constant pressure was delivered from the inverted bottle A. The arrangement for keeping the pressure constant will he readily understood from the figure .
Since the air was delivered at constant pressure , the rate of bubbling was constant , and so the total number of air-bubbles could be estimated by determining the total time and the number of bubbles delivered during a short interval of time ( five minutes ) .
The constancy of the rate of bubbling was frequently checked during an experiment .
The bubbles were formed at the jet B , which was made by sealing a piece of Fig. 3 .
wider tubing to a piece of capillary tubing ( or blowing a small elongated bulb at the end of the capillary tubing ) and then heating and pulling this out to a fine conical point , the end of which was then snipped off .
By means of this form of jet and the constricted capillary piece C , the bubbles formed regularly at the jet , " hanging " for a maximum period of five seconds before being dislodged .
After leaving the jet the bubbles had to traverse the glass spiral showm in the figure , taking about 20 seconds to do so .
By means of these devices the change of concentration due to surface adsorption of the solute was localised 1911 .
] dynamical Theory of Interfacial Concentration .
565 as far as possible in the lowest compartment of the " fractionating " column D. The latter , as shown in the figure , was divided into a number of compartments by sealed-in inverted funnels , whose stems were alternately bent in opposite directions .
That this arrangement had the desired effect , as explained above , was demonstrated by placing some KMnC\gt ; 4 in the lowest compartment and bubbling air through for 20 hours .
Even after this period the permanganate had not found its way into the upper compartments .
The bubbles finally make their way into the wide vessel E , slide along the glass plate shown , and finally collect and coalesce at the surface of the liquid in E at places as far removed as possible from the point vertically above the exit tube of the fractionating column .
In this way the excess of the solute produced by the breaking or coalescence of the bubbles at the surface of the liquid in E is prevented from finding its way back into the fractionating column .
By analysing the total liquid contained in the first six compartments of the fractionator , we may therefore be sure of determining the total loss of solute due to the excess carried at the surfaces of the bubbles .
In the earlier experiments the column D was water-jacketed to minimise mixing effects due to heat convection .
Parallel experiments showed , however , that it was sufficient to keep the apparatus in a room of moderately constant temperature .
The volume of air delivered in an experiment was determined as follows :\#151 ; When ready to start , A was completely emptied of water , connection was made with C by rubber pressure tubing , which was then compressed by a screw-clip , and water allowed to flow into A from the reservoir F until equilibrium was established .
The experiment was then started .
After completion , A was cut off from connection with C and F , and the contained water run out and measured .
A was then closed to the air , water run in from F until pressure equilibrium was established with the same ( constant ) pressure-head as before , and then this water was run out and measured .
The difference of the two volumes of water gives the volume of air displaced from A under the constant pressure-head .
This volume had to be corrected to that corresponding to the hydrostatic pressure at the jet .
A further correction is required for the expansion of the bubbles during their ascent .
In the actual adsorption experiments the liquid column above the jet measured 82 cm .
, corresponding to 60 mm. of mercury , and an expansion of the bubbles of about 8 per cent , of their volume .
A correction for one-half of this was employed .
No correction was applied for the capillary pressure in the bubbles , since , their average radius amounting to about 1 mm. , the capillary pressure can be readily shown to amount to less than 0'2 per cent , of the total hydrostatic pressure at the level of the jet .
566 Prof. Donnan and Dr. Barker .
Gibbs ' Thermo- [ July 31 , Prom the volume of air thus determined and the number of air bubbles , the total adsorbing area is readily calculated .
The apparatus was filled and emptied as follows:\#151 ; To fill it before an experiment , air pressure is applied through the side tube of the flask G , whereby the solution is driven from G into D and E. The tap H is then closed .
During the filling air must be kept constantly bubbling from the jet , for , if this be not done , the solution gets into the capillary tubing and sets up a high capillary pressure .
After the experiment is over the solution is syphoned off from E , and then the liquid in the six lower compartments of D removed by means of the side tube , measured , and analysed ( by means of the dropping-pipette ) .
The volume of solution so analysed wTas always 255 c.c. The lower end of D was closed by a paraffined cork , through which the-tube of the air-jet passed .
It was found that this paraffined cork removed a little nonylic acid from a fresh 'solution , when the latter was allowed tu stand in contact with the cork for 24 hours .
A state of equilibrium was , however , attained , for the cork when so treated did not now affect another solution of nonylic acid of the same concentration .
The precaution was , therefore , always taken of " saturating " the cork before an experiment was carried out .
The validity of this procedure was amply demonstrated .
A further possible source of error was the vaporisation of the nonylic acid into the bubbles .
It was therefore necessary to investigate by blank experiments whether any appreciable effect of this sort occurred .
Air was-bubbled continuously for 24 hours through solutions of nonylic acid contained in D , the top vessel E being removed .
In this case the bubbles broke at the surface of the solution situated in the upper portion of the adsorption-vessel D. The solution was analysed before and after an experiment by means of the dropping-pipette .
The following results show that the vaporisation loss was extremely small:\#151 ; Strength of solution .
Drop-number before .
Drop-number after .
per cent. 0 -00243 349 348 0-005 410 *4 410 -1 0 -00806 482 481 6 .
Calculation of the Measured Adsorption.\#151 ; The method of calculating the-adsorption from the experimental data will be made clear by the following example :\#151 ; 1911 .
] dynamical Theory of Interfacial Concentration .
Data.\#151 ; Barometric pressure ... ... ... ... ... ... ... ... Pressure-head in Reservoir A ... ... ... ... ... " at air jet ... ... ... ... ... ... ... . .
Volume of air driven over from A ( at pressure in A ) ... ... ... ... ... ... ... ... ... ... ... ... Temperature of experiment ... ... ... ... ... ... Number of bubbles per minute ... ... ... ... ... Duration of bubbling ... ... ... ... ... ... ... . .
Drop number of solution before experiment ... \#187 ; t\gt ; after ^ ... The total volume of air in the bubbles , calculated for half the height of the pressurehead in D ... ... ... ... ... ... ... ... ... ... . .
Total number of bubbles ... ... ... ... ... ... .
Hence the average diameter of a bubble ... .
and the total adsorbing area ... ... ... ... ... 728 mm. 125 " 4 cm .
water = 92*3 mm. mercury .
82 " = 60*3 " 90'5 c.c. 15'\#151 ; 16 ' .
24 .
20*6 hours .
482-7 .
479-3 .
90-5 x = 97-9 c.c. 758-2 24 x 60 x 20-6 = 29640 .
1*848 mm. , 3179 sq .
cm .
With the drop-number and surface-tension curves given previously it can be calculated from the above data that the decrease of concentration of the nonylic acid amounted to-0-0001099 per cent. , i.e. , 0*0001099 grm. per 100 c.c. Hence the total amount of nonylic acid removed by surface-concentration ( adsorption ) at the total air-liquid interface Whence follows 0 0001099 w a\#151 ; n.rwK\gt ; QA = \#151 ; --------x 255 = 0*000280 grm. " 0-000280 n.QQ 1 A\#151 ; r / , r = ~3n9~~ = 0 88 X 10^ grm./ cm.2 .
The experimental data are given in the following Tables .
The volumes given refer to the total volume of the air-bubbles , as corrected for the-pressure-difference in the manner explained above .
The change of concentration produced by the adsorption can be calculated in each case from the direct experimental data ( determinations of drop-number before and after the experiment ) by a reference tu the curves given previously .
Table II.\#151 ; 0'00243-per-cent .
Solution of Nonylic Acid .
Volume of air .
Bubbles Adsorbing area .
Drop-number .
Temp. per 1 minute .
Time .
Before expt .
After expt .
r x 107 .
o o c.c. hours .
cm.2 .
grm./ cmA 14 -7\#151 ; 15 -5 209 31-4 .
22-2 5920 348-6 344-7 0*93 14 -0\#151 ; 15 -7 272 41 *4 19 *3 7390 349 -2 343 -9 1 -02 13 -4\#151 ; 14 -2 78 94*3 1 24 *65 4574 346 -1 343 -2 0*90 Mean ... 0'95 ( 568 Prof. Donnan and Dr. Barker .
Gibbs Thermo- [ July 315 Table III.\#151 ; 000500-per-cent .
Solution of Nonylic Acid .
Temp. Yolume of air .
Bubbles per minute .
Time .
Adsorbing area .
Drop-number .
r x 10b Before expt .
After expt .
o o c.c. hours .
cm.2 .
grm./ cm.2 .
16 -0\#151 ; 17 '0 86 '5 16 '4 16 '4 1904 408 '3 404-9 1 '71 14 -0\#151 ; 15 '0 105 -5 5 '0 41 3 2554 408 -4 405 -8 1 -39 !
12 -0\#151 ; 13 -0 93 '7 14-0 19 -8 2550 409 '5 406 -5 1 '25 | 13'5\#151 ; 14-3 106 '1 22 -6 19 '2 3210 414 '0 409 -25 1 -65 i 15'0 48 -0 47 '0 18 '7 2387 414 -7 411 -0 1 '71 j 13'8\#151 ; 14-8 47 '0 47 '6 19 '3 2400 414 '5 411 -2 1 -52 Mean ... 1 -52 Table IV.\#151 ; 0'00759-per-cent .
Solution of Nonylic Acid .
j Temp. Volume of air .
Bubbles per minute .
Time .
Adsorbing area .
Drop-number .
r x ioh Before expt .
After expt .
O O 13 -9\#151 ; 14 -4 14'4\#151 ; 15'3 c.c. 186 '0 190 '4 13 -5 12 '0 hours .
21 -7 26 '3 cm.2 .
4098 4262 471 -7 466 -25 467 '0 460 '0 grm./ cm.2 .
0'95 1 -23 Mean ... 1 '09 Table V.\#151 ; 0'00806-per-cent .
Solution of Nonylic Acid .
Temp. Volume of air .
Bubbles per minute .
Time .
Adsorbing area .
Drop-number .
r x lo7 .
Before expt .
After expt .
o o 13 '0\#151 ; 14 -0 13 -0\#151 ; 14 -0 14 -0\#151 ; 16 '0 15 -0\#151 ; 16 '0 c.c. 121 -0 230 -5 97 '9 79 '5 9 9 9 9 O O tP i\gt ; I ( M CO ( M H j\#151 ; 1 hours .
18 -5 23 '5 20 -6 17 *5 cm.2 .
3322 6373 3180 4440 483 -8 483 '8 482 -7 482 '1 479 '5 477 '6 479 '3 477 '5 grm./ cm.2 .
1 '10 0*81 0'88 0'87 Mean ... 0 -915 It will be seen from the above figures that the values determined for each solution show a very satisfactory agreement amongst themselves , in spite of large variations in the speed of bubbling .
It seems very probable , therefore , that the measured adsorptions represent the true equilibrium values .
The mean results given in Tables II to V show a maximum adsorption occurring 1911.1 dynamical Theory of Interfacial Concentration .
in the neighbourhood of 0'005-per-cent .
concentration .
The mean value of T , 1*5 x 10-7 grm. per square centimetre , obtained for the 0'005-per-cent .
solution , , deviates from the other mean values by an amount which , appears to exceed the experimental error , but the latter values themselves lie pretty close-together .
More experimental work will be necessary before the existence-of such a maximum is established beyond doubt .
7 .
Calculation of Y from the ( cr , c ) Curve.\#151 ; In calculating T from the- equation T = \#151 ; ^ , E = 2 x 4'2 x 107 ergs per gramme-molecular weight , if the gramme-molecular weight be taken as the unit of mass .
Since the molecular weight of nonylic acid is 156 , we must put E = 2 x 4-2 x 107 -r-158 ergs per gramme , if we take the gramme as the unit of mass .
In that case the equation will give Y in terms of grammes per square centimetre .
This assumes that the nonylic acid is completely undissociated .
Taking the electrolytic dissociation into account , we must write Y = \#151 ; -r-+ .
\ tXv X d/ C where i is va n't Hoff 's factor .
Assuming complete electrolytic dissociation of the nonylic acid in the extremely dilute solutions employed , we have-i = 2 .
The values of dafdc corresponding to the various values of c may be-calculated from the ( \lt ; r , c ) curve obtained by the authors ( see p. 560 ) , or from Forch 's curve.* The values of Y calculated in this way , in ten millionths of a gramme per square centimetre , are shown in Table VI .
For convenience of comparison , the experimentally determined values are included in the table .
Table VI.\#151 ; Comparison of Observed and Calculated Values of R Concentration .
rx 107 observed . .
r x 107 calculated\#151 ; From Forch 's curve ( 18')\#151 ; From our curve ( 16 *5')\#151 ; For i \#151 ; 1 .
For i \#151 ; 2 .
For i = 1 .
For i \#151 ; 2 .
per cent. 0 *00243 0-95 0*58 0-29 0-55 0 -26 0 -00500 1 *52 1 23 0-61 1 -14 0-57 0 *00759 1 -09 1-58 0-79 1 *26 0*63 0 *00806 i 0 915 1 -63 0*81 The calculated values of F increase with the concentration , and do nob show any maximum .
* Forch , loc. cit. 570 Prof. Dorman and Dr. Barker .
Gibbs - [ July 31 , It will be seen that the calculated and observed values of T agree as to order of magnitude .
For i\#151 ; 2 the agreement between the observed and calculated values is better at the higher concentrations .
Considering the difficulty of measuring such small changes of concentration in the extremely dilute solutions employed , the agreement must be regarded as affording a verification of Gibbs ' fundamental equation .
8 .
The Adsorption of Saponine at an Air to Water Interface.\#151 ; An investigation of a similar kind to that carried out with nonylic acid was extended to the glucoside saponine .
As is well known , this substance forms very stable foams and viscous films at the bounding surface of air-bubbles .
This peculiar behaviour rendered an investigation of the surface adsorption in this case of considerable interest .
Ordinary saponine is unfortunately not a very definite substance , its composition varying with the source of supply .
The preparation used in the present work was Merck 's " pure white " preparation .
The mean molecular weight in aqueous solution was determined by freezing-point measurements , whereby moderately strong solutions were essential for obtaining an accurately measurable lowering of the freezing-point .
For solutions varying from 5 to 10 per cent. , the mean value so found was 1260 .
This value has been employed in the subsequent calculation of the adsorption in very dilute solutions , although this procedure is open to some objection .
We had , however , no means of determining the molecular weight in these very dilute solutions .
It was found necessary to work with fresh solutions , as the " drop- .
number " of a given solution , obtained with the same dropping-pipette , increases slowly with time , the increase becoming detectable with our apparatus in three to four days .
Furthermore , the drop-number was found to vary very considerably with the speed of dropping .
This is probably due to the high molecular weight and consequent slow rate of diffusion causing a relatively considerable time to be required for the establishment of the surface-concentration at the fresh air and water surface .
Thus for the same volume of solution in the pipette the number of drops diminishes as the speed of dropping increases .
As pointed out before in the case of nonylic acid ( where a similar , though not so marked , effect occurs ) , this source of error has no effect on the use of the dropping-pipette as a method of analysis , provided one works under strictly constant conditions .
The following data were obtained with a constant total time of dropping ( 50 minutes ) .
Temperature = 16-5 ' .
1911 .
] dynamical Theory of Interfacial Concentration .
571 Table VII .
Concentration of saponine .
Drop-number .
Surface-tension .
per cent. 1 1 dynes/ cm .
0 324*6 72-91 0-0072 333 -9 70-88 0 0143 346*9 68-42 0*0320 351 -3 67-37 0 0433 353 -8 66-89 The data of Table VII have been used for determining the change of concentration of the saponine solution employed in the experiments on adsorption .
The corresponding surface-tensions are shown in Curve I of fig. 4 .
In order to ascertain the effect of variation of speed of dropping on the values of the surface-tension determined in this way , the following experiments were performed:\#151 ; Table VIII .
Drops per minute .
No. of drops .
Concentration = 0-00708 per cent. 6- 66 333 -3 460 338-0 3*55 341 -9 2-60 344-0 2-50 344-3 Concentration = 0-01413 per cent. 6-90 346 -5 4-77 349-2 3-66 351 -5 2-59 353 0 Concentration = 0-032 per cent. 7*03 351 -3 4-43 354 -6 2*62 357 -7 Interpolating for the constant very slow speed of dropping of 2*6 drops per minute , the following results are obtained ( temp. = 16*5'):\#151 ; Table IX .
Concentration of saponine ( c ) .
Surface-tension ( \lt ; r ) .
per cent. dynes/ cm .
0 72 -91 0 *00708 68 -80 0-01413 67-04 0 032 66-16 57 2 Prof. Donnan and Dr. Barker .
Gibbs ' Thermo- [ July 31 , The data of Table IX are plotted on Curve II in fig. 4 .
It will be seen that in the initial part Curve II is considerably steeper than Curve I. XOO 200 300 400 Concentration of Saponine ( in ben-thousandths of x per cent ) Fig. 4 .
The adsorption was experimentally measured in the same manner as before , the results being given in the following table ( temp. \#151 ; 15'):\#151 ; Table X.\#151 ; Concentration of Saponine = 00072 per cent. Volume of air .
Bubbles Adsorbing area .
Drop-number .
per minute .
Time .
Before expt .
After expt .
r x io7 .
c.o. hours .
cm.2 .
grms./ cm.2 .
141 -8 126 22 -7 7290 336 -7 334 -0 5 *2 205 -1 147 24 -8 10155 333 -6 331 -8 2*6 263 '6 126 .
24 -3 11313 334 -4 332 -0 3 *1 Mean ... 3 *7 In order to calculate T in this case from the equation T c 7RT ' dc* we have 4 = 1 and E = 8'4 x 107/ 1260 ergs per gramme .
The results of this calculation and the experimentally determined adsorption for the same concentration ( 0'0072 per cent. ) are as shown below :\#151 ; r x io7 observed .
P x 107 calculated .
\#151 ; from Curve I. dc \#151 ; from Curve II .
dc 3-6 1 -36 1 -60 1911.1 dynamical Theory of Interfacial Concentration .
573 The observed value is about double the calculated value , but , considering the difficulties and uncertainties of the determinations , the agreement is not unsatisfactory .
It would appear , therefore , that no irreversible " gelatinisa-tion " occurs in the surface-layer of an aqueous saponine solution , at all events in the case of moderately fresh solutions .
Possibly the " gelatinous skins " which may form on standing are due to secondary causes , or at all events to slow irreversible effects .
9 .
Summary of Results.\#151 ; ( 1 ) The surface-tensions and surface-concentrations at an air to water interface have been experimentally determined in the cases of solutions of nonylic ( pelargonic ) acid and saponine .
( 2 ) In the case of very dilute solutions the adsorption ( surface-concentration ) was found to be of the order of 1 x 10"7 grm. per square centimetre for nonylic acid , and of the order of 4 x 10-7 grm. per square centimetre for saponine .
Working with solutions of sodium glycocholate , Congo red , and methyl orange , Lewis ( loc. cit. ) found the adsorption at an oil to water interface to amount to about 5 x 10-6 grm. per square centimetre .
( 3 ) The values of the adsorption found for nonylic acid and saponine agree as to order of magnitude with the values calculated from Gibbs ' equation .
Considering the difficulties and possible errors of the experimental determinations , the agreement between the observed and calculated values may be regarded as a satisfactory verification of the validity of Gibbs ' theory of surface-concentration for an air to water interface .
VOL. LXXXV.\#151 ; A.

