]\gt ; [ 101 ] III .
Tonic Size in Relation to the Physical Properties W. R. by Professor LARMOR , , R. Received January 5 , \mdash ; Read February 8 , 1906 .
PART I.\mdash ; INTRODUCTION .
PART ExPERIMENTAL DETERMINATIONS .
Pagc ( a ) Density and viscosity of solutions at C. . . . . . . . .
107 ( b ) Density , viscosity and conductivity of mixtures of normal solntions of KC1 C. . . . . .
111 ( c ) Density of KC1 and solutions at C. . . . . .
113 PART III.\mdash ; IONIC SIZE 1N ATION To CoNDUCTI VITY .
( a ) Theoretical considerations . . . . . .
114 ( b ) Deduction of constant for . . . . . .
117 ( c ) Distinction between autolytes and heterolytes . . . . . .
119 ( d ) Mixtures .
Density and conductivity . . . . .
120 ( e ) Hittorf numbers a function of the dions . . . . .
121 ( f ) Deduction of hydration numbers and values of radions and ionic volumes . . .
124 PART IONIC SIZE 1N RELATION To DENSITY .
( a ) Solution volume a linear function of ionic volume 126 ( b ) Deduction of density formulae . . . .
128 PART V.\mdash ; IONIC SIZE 1N RELATION To VISCOStTY .
( a ) Viscosity a function of the radions . . . . . . . . .
129 ( b ) References to previous theories . . . . . . . . .
132 considerations leaaing to a general viscosity formla . . .
134 ( 404 .
) 7.6.06 Page ( d ) Extended conception of the radion . . . .
135 ( e ) The general viscosity formula . . . . .
135 ( f ) Experimental tion of u formula 136 ( g ) Comparison with results 141 PART IONIC SIZE 1N RELATION To FREEZING-P01NT DEPRESSION .
( u ) General considerations . . . . . . . .
142 Effective molecular freezing-point depression a linear function of ionic volume 143 ( c ) Effective molecular freezing-point depression a linear function of solution volume 146 Correction of freezing-point depression by reference to ionic volumes . . . . .
148 PART IONIC SIZE 1N RELATION To REFRACTIVE POWER .
PART VIII .
CONCLUDING OBSERVATIONS .
APPENDIX .
Note on the Van Hoff . . . .
155 PART I.\mdash ; INTRODUCTION .
THE electrical conductivity of a solution depends upon the rates of transference of the ionised portions of the solute in opposite directions under the influence of the applied electro-motive force .
These rates of transference under a given potential gradient are conditioned by the viscosity of the medium and the sizes and possibly the shapes of the ions .
Increase of viscosity of the solution and increase in the sizes of the migrating ions both tend to diminish the rates of transference of the ions , and thus to lower the conductivity .
If the ion enters into combination with one or more molecules of water , its size is necessarily increased , and the motion of the waterlogged ion more sluggish as the amount of water in combination increases .
To separate the elements which determi .
no the conductivity of an electrolytic solution , and to analyse the joint effect of variations in ionisation , viscosity and water combination is a matter of great difficulty , but of much importance to the theory of solution .
In a former paper ( BOUSFIELD on " " Ionic Sizes in Relation to the Conductivity of Electrolytes\ldquo ; * ) was proposed a method for effecting such an analysis based upon the expression evaluated by STOKES for the terminal velocity of a small sphere moving in a viscous medium .
A consideration of the influence of the water in combination with the ion upon its mobility was used to obtain a correction of the coefficient of * Communicated to the Royal Society , February 10 , 1905 .
Revised and published in the ' Zeit .
fur Phys. Chem 53 , 257 , 1905 .
( See infra , Part VIII .
, p. 153 .
) THE PHYSICAL RTIES OF AQU OUS SOLUTIONS .
ionisation , which made VAN HOFF'S law ( in a slightly modified form ) an ) expression of the relation between ionisation and dilution , down to twice decinormal solutions of KC1 .
This method of procedure gave for the radius of the hydrated ion an expression of the form which indicated that the average radius of the ion steadily increased with dilution , owing to increasing hydration of the ion , up to " " infinite dilution Whether the resulting expression for did in fact represent the average radius of the ion was tested by a consideration of the density law which would thence result .
The volumes of the ions would be proportional to , and it was found that a rational density formula could be constructed upon this basis which accurately esponded with the observed densities of the solutions .
In the present paper the necessary experimental determinations for applying the method to solutions of are given , together with some other determinations for collateral purposes , and the hypothesis is further tested by ence to other physical properties of solutions .
These results are independently of some value , apart from the hypothesis by which they are reached , and it has been thought well to designate the function , which , according to our view , expresses the average radius of the ion , by the term " " radion Whilst this term connotes our hypothesis , it may , if necessary , be merely considered as expressing a certain function of the dilution .
But it has a further convenience , as it enables us readily to extend the conception to denote the average molecular radius of any group of ions or molecules , or even of the whole of the ions and molecules both of solvent and solute in any given solution .
The utility of this extended conception will appear more clearly in the section relating to a consideration of ionic size with reference to the viscosity of a solution .
The volume of the ion , according to our hypothesis , is proportional to the cube of the radion , and the volume of a pair of to the sum of the cubes .
These cubes and their sums we refer to as " " ionic volumes But except where the context indicates the contrary , the term " " radion\ldquo ; may be taken as merely denoting the function and the term ionic volume\ldquo ; as denoting the cube of the adion or sum of the cubes of the radions , apart from the hypothesis as to size .
As in our former paper , with reference to , it is shown that the " " solution volume\ldquo ; of solutions is a linear function of the ionic volumes .
Hence the densities of KC1 and solutions can both be accurately expressed by the same formula as simple functions of the radions .
A theoretical consideration of the relation of the Hittorf gration numbers to the sizes of the ions is given , and it is shown that our hypothesis as to the influence of ionic sizes upon rates of transference would theoretically lead to the result that the reciprocals of the Hittorf migration numbers should be expressible as a linear function of the ratio of radions .
This turns out in fact to be the case , and we incidentally arrive at a useful method of extrapolation to determine the value of the Hittorf number for an electrolyte at " " infinite\ldquo ; dilution .
These considerations enable us to determine the coefficients for the separate ions in the expressions for the radions .
These coefficients we refer to as the " " hydration numbers the relation between the ation numbers and the migration numbers at infinite dilution being of the form Turning now to a consideration of the viscosities of the KC1 and solutions , it is shown that the viscosity of these solutions can be represented approximately as a linear function of the radions , as can also the viscosity of mixtures of normal KC1 and solutions .
Passing to a consideration of the general relation of viscosity to ionic size , the extended conception of the radion is introduced , and an approximate value is given to the radion of water , which expresses the average radius of the water molecules reckoned upon the same scale as the radions of the solute .
Using this value of the water radion in conjunction with the values of the radions of the solute determined from the conductivities , it is shown that the viscosity of the solutions can be expressed with a fair approach to accuracy by the expression where stands for the radion of a given species of molecules , and for the fraction of the total volume occupied by such species .
Since is , upon the extended conception of the radion , the average molecular radius of the whole solution , we may express this result by saying that the viscosity of an aqueous solution is proportional to its radion .
In order to correlate ionic sizes with osmotic pressure , a prolonged attempt was made to measure the vapour pressure of dilute KC1 solutions at C. A large barometer tube was used , closed by a small tap at the bottom , so that minute differences of level could be determined by remoying and weighing the mercury cistern .
A similar arrangement was used to determine simultaneously the variations of atmospheric pressure during each observation .
It was found , however , that the variations of atmospheric pressure were often larger than the differences of vapour pressure to be measured , and no sufficiently accurate results could be obtained .
Recourse was therefore had to the freezing-point determinations of JAHN with KC1 and solutions .
The val.iations of ionic size with temperature are probably serious THE PHySICAL PROPERTIES OF AQUEOUS SOLUTIONS .
and at present*unknown , and hence a consideration of ionic sizes at C. in relation to osmotic data at another temperature might lead to error .
But it seemed probable that the ionic sizes at different dilutions might have the same values at C. and C. , and in the absence of other data it was decided to use these .
Defining the " " effective molecular freezing-point depression\ldquo ; as the ordinary so-called molecular freezing-point depression divided by , where is the ionisation , and denoting it by the letter , it was found that was a linear function of the ionic volume , and that it could be expressed both for KC1 and as , where stands for the ionic volume at the given dilution and for the ionic volume at infinite dilution , the value of the constant being nearly the same for both substances .
In addition to the confirmation thus afforded to our view as to the fundamental importance of the radion in the theory of solutions , we are further led to a useful formula for obtaining by extrapolation the value of the molecular freezing-point depression at " " infinite\ldquo ; dilution .
For this purpose we are able to dispense with our hypothesis , and obtain from it a new result quite independent of it\mdash ; one of the recognised tests of the validity of a hypothesis , though not a conclusive one .
We saw that the solution volume was a linear function of the ionic volume , and we have now the effective molecular freezing point-depression also as a linear function of the ionic ] .
Hence the molecular freezing-point depression should be a linear function of the solution volume , and in this reference to ionic sizes which correlated the two sets of phenomena can be dispensed with .
In order to test this matter , measurements of the densities of KC1 and solutions at C. were made .
This is a little above the freezing point of dilute solutions , but it was considered to be near enough to make the desired comparison .
Density measurements were made upon solutions of strengths of normal , and empirical formulae ( based on the lines of the rational formulae for C. ) were constructed to obtain the solution volumes at the concentrations at which JAHN'S freezing-point determinations were made .
The result was that the eflective molecular freezing-point depression both for KC1 * In my former paper I attempted to calcul the variation of ionic size temperature by reference to the condnctivity } ) coefficients of the ions at infinite dilution which were given by KOHLRAUSCH .
I have , however , since come to the conc.lusion that these resnlts are u1lrelial)le .
values were largely based ou determinations of the collductivities of normal soIutions .
Dissociation being incomplete in solutions , any variation of conductivily due to change of ionisation with temperature ) included in his temperature coefficients and might entirely vitiate the deductions which I drew .
This portion of my former paper mnst therefore be propose to pursue the matter .
experimentally .
VOL. and solutions could be expressed by one formula with the same constants for both substances , .
:\mdash ; where is the change in the solution volume for different dilutions .
This formula would also include a non-electrolyte such as sugar , in which is almost zero .
Furthermore , by correcting the concentration for the amount of water abstracted by and combined with the solute , a constant value was obtained for the effective molecular freezing-point depression of KC1 solutions down to a concentration of twice decinormal .
The equivalent refractive power of KC1 and solutions was similarly shown to be a linear function of the ionic volume , and , consequently , also of the solution volume .
In the sketch of the course of the present paper we have passed over some matters arising incidentally which may call for mention .
But it must be obseryed that the main purpose kept in view throughout is to show the interpenetration of the theory of ionic sizes with the theories of the various phenomena of solutions , and to test the theory as far as possible in its relation to such phenomena .
In this process various side avenues have been opened up , which we have forborne to follow if they carried us too far from the main track .
One incidental matter of importance is the correction of the coefficient of ionisation which is afforded by the theory , according to which the true value of the coefFicient of ionisation is The result of this correction is to make the values of for KC1 and very nearly the same for equimolecular solutions , down to twice decinormal concentration .
In confirmation of this it was found that the obseryed densities of mixtures of normal solutions of KC1 and were correctly given by the " " law of mixtures\ldquo ; within very narrow limits .
The conductivities of these mixtures were also determined and a mixture law for the conductivities based upon the consideration of viscosities and ionic sizes was formulated .
The agreement thus obtained between observed and calculated values also tends to show that the hypothesis upon which the calculations were made is correct .
The incl.ease in the ionic volumes which takes place with increasing dilution must be nearly proportional to the increasing volume of water combined with the ion .
The development of this matter quantitatively , so as to determine the number of molecules of water in combination with the ion under different circumstances , is a matter of great importance , but it cannot be dealt with in this paper .
The effect of water THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
combination as a sufficient and possibly the only cause of ionisation is dealt with , to some extent , in the former paper ( see the revised version published in the ' Zeitschrift Phys. Chem .
cit. ) and has also recently been considered by LOWRY An Application to Electrolytes of the Hydrate Theory of Solution 'Trans .
Farad .
Soc. , ' I. , ] , 1905 .
PART II .
DETERMINATIONS .
Density Viscosity of Solutions at C.\mdash ; The material used for the earlier determinations was purest fused ( obtained from BAIRD and TATLOCK ) .
A stock solution of a strength of per cent. was prepared , the density of which at C. was found to be MENDEL ( ' Principles of Chemistry , ' ed. 1905 , vol. , p. 451 gives for the density of solutions at C. the formula water .
This formula gives for per cent. the density A further sample of material was ) urified by precipitation with HC1 , recrystallised and fused ( specially prepared by MERCK ) .
A stock solution of per cent. was made from this sample , the density of which at C. was .
MENDELE'EF'S formula gives .
No substantial difference between these solutions manifested itself , but for any range of values the same solution was used .
The densities were determined in all cases in duplicate , by means of a pair of yknometers , each having a capacity of about 100 cub. centims. The viscosities were determined for the most part by means of the simpler type of viscosimeter described in the former paper .
The results from this instrument are indicated in the diagram by black circles .
A few determinations which were made with another instrument are by open circles .
The viscosities were calculated from a differential formula derived from the formula .
By differentiating we have @ For water , ; and for the instrument chiefly used , , with a small correction for variations when the time of flow of water at C. varied from 621 seconds .
These constants were determined by the time of flo of water at various temperatures , as derived from THORPE and RODGERS ' determinations ( ' Phil. Trans vol. 185 , p. 397 , 1894 ) , by means of the formula given in a former paper ( BOUSFIELD and LowRY , ' Roy .
Soc. Proc vol. , p. 47 , 1902 ) , viz. :\mdash ; 08 MR. W. R. BOUSFIELD : IONIC SIZE IN RELATION TO The differential formula becomes on substituting for and S000008406 S. It was found by trial with a series of water values at different temperatures that the differential formula remained sufficiently accurate even when St amounted to as much as 50 seconds .
The values determined for the less dilute solutions of are given in the following table:\mdash ; TABLE I.\mdash ; Density and Viscosity of Solutions at C. Viscosity referred to water as unity .
The duplicate density determinations in no case differed by more than a unit in the fifth place of decimals .
The air corrections were determined from observation of barometer pressure and laboratory temperature , and it is considered that the densities may be taken to be relatively correct within in the fifth place of decimals .
The viscosity determinations have not the same order of accuracy .
For the purposes of this paper it was a matter of considerable importance to determine the viscosities of more dilute solutions , but this proved to be a matter of the greatest difficulty .
Other observers of viscosities have had the same experience , but whether owing to the same cause is not clear .
In the former paper it was pointed out that the constants of the apparatus , as tested by the time of flow of water at C. , vary from time to time in an irregular manner .
No amount of washing will bring the apparatus to a steady condition .
In dealing with very dilute solutions the variations of the water constant are often of the same order of maglNitude as the variations due to variation of concentration of the solution .
Hence the only safe method of operation is to interpose between the observations of salt solutions observations of water , which , of course , involves a thorough rinsing of the apparatus , which may in itself introduce some change in its condition .
When a series of observations of solutions with interposed observations of water shows no material change in the water constant , it is assumed that the conditions were the same throughout , and that the obSeryatiolR are therefore accordant .
But his assumption is not always true .
A very large number of observations on dilute THE RTIES OF AQUEOUbt SOLUTIONS .
solutions were made , many of which had to be rejected in whole or in part , on account of variations in the water constant .
Particulars of the best and most accordant set of observations on very dilute solutions are given in Table II .
* TABLE IT .
Water Time.flow .
In this series the mean value for the time of flow of water at at the beginning and end of the observations remains constant at .
At one point there was a variation in the water constant , but it returned afterwards to the same value , and * As my viscosity determinations for very dilute solutions differ materially from , which are given below , I have thought it well to set out fully the data from which my results are calculated , though very little use has been made of data for solutions below 1 per cent. strength in this )apel .
exc.ept for the initial process of correcting .
The effect of the solute in dilute solution upon the water itself , for which accurate measurements of dilute solutions reqnired , is left over for the time .
there appears to be no break in the values for the solutions .
St is therefore calculated throughout from the value for water of .
S is calculated from the density at 1 per cent. by a linear formula .
The series of observations set out in the two foregoing tables are plotted in Diagram I. , and shown by black circles .
A series of observations given in a recent paper by GRUNEISEN ( ' Wiss .
Abhand .
, Phys. Tecb .
Reichs 4 , 239 , 1905 ) , to which fuller reference will later be made , is also plotted .
They are indicated by double open circles , and the values are set out below .
III .
Determination of the Viscosity of Solutions at C. It will be obsel.ved that in the more concentrated solutions there is a fair agreement .
But at the dilute end of the curve the present observations show a considerably greater viscosity than GRUNEISEN'S observations .
It hardly seems possible that the difference can be due to changes in the apparatus constant during the series , as the water constant is the same at the beginning and end .
No other series so coherent as the one set out above was obtained , but the following observations , made with anocher instrument , tend to show that the difference is slightly greater than is indicated by the above series .
TABLE IV.\mdash ; Other Determinations of Viscosity of Dilute solutions .
These determinatio1ls have only the value of isolated observations , but they are entitled to some weight , and they are indicated on the diagram by small open circles .
The curve drawn on the diagram indicates the line of values actually adopted for the THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
purpose of this paper .
It was drawn on a larger scale for the very dilute end , and the values adopted for the different concentrations for very dilute solutions were taken from this curve .
3 4 5 6 7 .8 9 Diagram I. Viscosity of Solutions at C. ( b ) , Viscosity and Conductivity Mixtures Vormed Sof KC1 .\mdash ; In this series of experiments , stock normal solutions of KC1 and were made up , and the density , viscosity and conductivity of these solutions and of mixtures thereof were determined at C. The mixtures were 1nade by means of 100- and 200-cub .
centim. flasks ( stamped at the P.T.R. ) .
The densities and viscosities were determined as before described .
The conductivities were measured by KOHLRAUSCH'S method with wheel bridge and telephone with the special variations described in a former paper ( BOUISFIELD and LowRY , 'Phil .
Trans vol. , p. 286 , 1905 ) .
The platinum pot containing a glass-shielded electrode was used as the conductivity vessel\mdash ; resistance capacity .
The conductivities observed are given in Table X. , Part III .
The duplicate density mea , surements corrected for differences of laboratory temperature and barometric pressure were as follows:\mdash ; .
\mdash ; Densities at C. of Normal KC1 and Solutions and Mixtures ther- The viscosity measurenlents were made in two groups , with an interval of three days between .
The water constant appears to be steady in each set , but differs by a second for the two sets .
The order of accuracy obtained is about per thousand .
.\mdash ; Times of Flow of KC1 and N Solutions at C. { PHYSICAL OF 11 .
) ) The viscosities are calculated from the formula , whel for the first series and for the second series the constants being corrected for the of the time of flow of water at 1 in the manner before indicated .
( r ) of of dilute.{:olutions of C.\mdash ; It was necessary for our purpose to determine accurately the solution volunnes of dilute solutions of KC1 and at the freezing-point .
A pair of pyknometers of large capacity ( about 330 cub. centims. ) was therefore used for this purpose .
The solutions were made up by dilution at C. normal solutions , and to .
The pyknometers were packed in ice in a refrigerator for two or three hours in order to secure a correct temperature adjustment .
The air corrections for the weighing operation were calculated from the temperature and pressure in the laboratory from time to time .
The observed densities are given to the sixth place of decimals , though the sixth place is only of value as fixing the fifth place more accurately .
The " " observed\ldquo ; solution volumes are calculated from the densities ) means of the formula , where is the density of water at C. , , is , concentration in gram-molecuIes per litre of solution , is the equivalent weight of the solute .
In order to obtain expressions the solution volume in terlns of the hydration a formula of the same form as is shown to hold for C. was used ( uide post , Part IV .
) , viz. , , where For densities at C. we were able to determine from the conductivities by the method stated in Part III .
( g ) .
But no sufficiently accurate serics of conductivity values at C. was available for determining in this manner .
Hence the value of was determined by taking trial values of until , on setting out the resulting values of V*upon the resulting value of , the points ranged approximately in a straight line .
A considerable difference in the value of makes little difference in the values of , or in the limiting value which is obtained by putting The value of was taken as 6 in the case of , and 4 in the case of , and with these values of the values of were calculated .
The values of ( the number of molecules of water per molecule of solute ) are not VOL. COVL quite the solutions of KC1 and of equal concentration , but this again makes no appreciable difference , and the mean values of were taken for both .
The empirical formulae thus derived for the solution volumes were for KC1 for In the following table are set out the observed densities , the " " observed\ldquo ; values of the solution volumes ( calculated from the observed densities ) , the values of and the solution volumes calculated from the above formulae and the differences .
The values of the solution volumes at infinite dilution ( obtained from the above formula putting ) are for , for TABLh1 .\mdash ; Densities and Solution Volumes of KCI and at C. PART III.\mdash ; IONIC SIZE 1N ELATION To CONDUCTIVITY .
Theoretical Consideration , ?
.\mdash ; Some confusion still exists in the use of terms employed in relation to the theory of conductivity , which makes it desirable to state the fundamental equations with precision , and to define the terms used with exactness .
It will be convenient to set out together the meanings of the symbols employed .
absolute velocity of kation ( expressed in centimetres per second ) under a potential gradient of 1 volt per centimetre .
absolute velocity of anion .
quantity of electricity liberated by 1 gramme ivalent of any monad ( coulombs ) .
concentration of solute in gramme equivalents per litre .
hydratio ( number of molecules of water per molecule of solute ) .
the ions , . .
.- .
THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
specific conductivity of solution reckoned in units such that esistance in ohms of a centimetre cube of the solution between its opposite sides .
A , the " " equivalent lductivity equivalent conductivity at " " infinite\ldquo ; dilution .
, so that true coefficient of ionisation .
'radius of an ion .
radius of ion at infinite dilutiou .
viscosity in absolute units .
Hittorf migration number for anion ; onfining ourselves to the case of binary electrolytes composed of lent ions , ( 1 ) , . . . . . .
( 2 ) , nd at infinite dilution . . . . .
( 3 ) .
It is orta n to note the true " " mobilities\ldquo ; of the ions , being to the velocities .
The numbel .
S usually tabulated as " " mobilities\ldquo ; and , whose sum makes up .
At infinite dilution the distinction ceases to exist , since , but it is important to keep it in mind for less dilute solutions .
According to STOKES ' theorem ( vide former , the limiting velocity of a small sphere of radius in a medium of viscosity under an applied force is proportional to .
Assuming that the ions are spherical and that they are water-coated , so that the friction is always water on water , we should have , if STOKES ' theorem is applicable to ionic movements under the electromotive force , . . .
' .
( 4 ) , where is a constant monovalent ions .
Assuming the validity of the application , the sizes uf the ions can ined in absolute units , as is done in Professol ' POYNTING'S calculation in ence to the former paper .
But for our present purpose we are concerned ttions derived from viscosity measurements of divalent ions doubtful whethol tho value of could be applied to them .
It may be that ions differ from a sphelical shape .
It is possible that the and OH ions , being little , if at all , hydrnted , might equiro ( diffel.eut contant .
I set out this culation in the paper published in the ' fiir ) , 1905 , p. 302 .
MR. W. , BOU FIELD : IONIC SIZE IN RELATION TO sizes , and are liberty to simplify the equations by choosing a suitable unit radius .
This we do by making in equation ( 4 ) , so that we have always or ( 5 ) , or , at infinite dilution , lu other , our unit radius is the radius of an imaginary watercoated ion which has unit mobility at infinite dilution .
All ionic radii will be given in ionic units , as we may shortly term this unit .
Moreover , for reasons which will more fully appear hereafter , we shall speak of the ionic radius as the " " radion This will be specially convenient , first , because the radion will be shown to be an important physical quantity , quite apart from our theory that it represents the adius of the hydrated ion ; secondly , because , as was shown in the former paper , the value of the dius which we obtain is the value for the whole of the ions , and not the value for any individual ion ; and , thirdly , because the conception of the adion is useful , not only in reference to ions , but also to other hydrated molecules , and even to water itself Slightly modifying the notation given in the former paper , we have , and writing where and written as abbreviations for and , so that , get for the true value of .
( 6 ) , , where ( 7 ) .
the case of it was shown that was of the form ) .
It will be hereafter shown that a function of the same form can be applied with a close approximation to the truth to the case of .
Assuming that and are of the same form as , we get . . . . .
( 8 ) , when and are the constants for the separate ions , and we get for the radions the expressions , . . .
( 9 ) .
lt be to des , \amp ; c. , as the ation n THE PHYSICAL RTIES OF AQUEOUS SOLUTIONS .
By determination of the quantities of each kind of ion which pass in opposite directions a given section of a solution in a certain time under the influence of a iven current , HITTORF and others have obtained various series of values known as ' ' migration numbers These are tabulated the lnions , often ) esellted by the lettel , so that In ordel to make this fit in ] with the present uomenclatul.e , it will be convenient to designate the Hittorf nulnber by and the COl.responding number for the kation by .
Then we have for the values of at infinite dilution Then , ironl equations ( 5 ) , ( 6 ) , and ( 7 ) , we get or ( 10 ) .
In the case of KC1 , equation ( 8 ) may be wlitte in th fornl . . . . . . .
( 11 ) , which is the relation between the hydration numbers and the tion nubers , and equation ( 10 ) gives us . . . . .
( 12 ) .
There arises thus from our theory a simple linear relation between the series of Hittorf numbers .
for electrolyte and the corresponding series of values of .
We shall see that this in fact holds , and that equation ( 12 ) gives us ready means for obtaining , the value of the Hittorf numbers at infinite dilutio1l , quantity of great importance in the theory of electrolytes .
( b ) Deduction of Constant for .\mdash ; In the former paper it was shown that in the case of KC1 the function was of the form , and that the value of was .
This function was evaluated for KO1 upon the hypothesis that the modified Van Hoff equation , . . . . .
( 13 ) , old it coefficient of were ected for viscosity and variations of ionic The coefficient of ionisation thus corrected according to our theory takes the form given above in equation ( 7 ) .
In the case of it is now found that a similar accord with the law expressed in equation ( 1S ) be obtained with function of the same form and a value of .
This is shown in the following table in which are given corresponding values of , and the other magnitudes involved .
The values of are derived from scale diagrams of the observed values .
For obtaining the values of there is taken the series of values of given by AUSCH ( ' Sitz .
der K. Preuss .
Akad .
der Wiss .
zu Berlin , ' 1900 , p. 1002 ) , .
The corrected values of are calculated by means of equation ( 7 ) with the value for .
The column headed gives value of the constant in equation ( 13 ) as calculated from these data .
The column headed shows the difference of from the mean value ( which is arrived at without reference to the value for , for the reasons given in the former paper ) .
TABLE VIII .
The colunm headed shows the difference between the observed and calculated value of , worked out ( as in the former paper ) from .
The differences are somewhat more uneven than in the case of , owing partly to the fact that the viscosity COll.ection is larger and probably less accurate , and partly to the fact that the function does not quite so well represent the correction .
But the accurate correspondence of the experimental values with the modified Van Hoff law is in this case brought down to normal solutions , within the limits of experimental error ; and the divergence is nowhere so great as one per thousand except on the extreme figul.e .
The following table shows the values of for KC1 and calculated in the ordinary way and the values of calculated from equation ( 7 ) or equation ( 13 ) .
* As to this hypothesis , vide Appendix , p. 155 .
Referred to vater as unity .
TH PHySICAL OF SOLUTIONS .
It will be seen that whilst the ordinary values of widely diverge at a concentration of twice decinormal , the corrected values of are nearly the same .
yond this point the values diverge .
TABLE IX .
\mdash ; Values of A greater divergenc in more concentrated tions is to be expected , and it is important to point out the reason therefore .
( c ) Distinction Hcterolytcs.\mdash ; In a recent paper ( BOUSFIELD and LowRY , 'Phil .
TralR , ' vol. , p. 256 , 1905 ) attention was drawn to the difference between autolytic and heterolytic conductivity , and it was shown that an electrolyte , such as sodium hydroxide , which was a good conductor in the melted state , showed clear indications of autolytic conductivity in the more concentrated solutions .
It may be supposed that , as in dilute solution , water combination is the effective force producing ionisation , so in the melted state self-ionisation in autolytic conduction ) is produced by the combination of the ions with molecules of the melted salt .
In concentrated tion , this action probably still persists , and though we have mainly heterolytic ionisation produced by tion of ions with the solvent , we may have also some autolytic ionisation produced by combination of the ions with molecules of the solute .
KC1 and belong to the class of autolytes , * Since this paper was communicated , I have made an examination of a group of nine salts with referencc to this matter , and have come to the conclusion that no stress should ) laid upon the near approach to cquality of the values of for KC1 and .
If water coml ) ination ) the efficient cnuse of ionisation , one wonld oxpect that the great differences in the amount of wnter combination with different wonld lead to corresponding differences in the of .
This tnrns to roI the These words were dopted by Dr. myself the of Professor ] to the to hich attcntion was This theory has ) recently developed paper by Society , ' vol. 1 , p. 197 , " " to Electrolytes of the Hydr .
eory of ltionS , ftnd if we took a concentrated solution of either salt at a suitable pressure and temperature above the melting-point of the salt , and were ] to drive off the water , we should come gradually to a condition of things in which the conduction was entirely autolytic .
Hence we cannot expect the near equality of ionisation which find in the case of and at the dilute end of the range to persist at the concentrated end where the idiosyncrasies of self-ionisation may produce a marked effect .
( d ) Mixtures of KC1 Normal Solutions.\mdash ; If two solutions have the sanue ionisation , there should be no change of molecular aggregation upon mixture , since , according to our theory , change of ionisation and change of ionic hydration .are necessarily concomitant .
Density measurements afford a very sensitive test of changes of molecular aggregation , and the results in this instance show that the change of aggregation upon mixing normal KC1 and solutions in various proportions is very small .
The question of the conductivities of such mixtures has a bearing upon the validity of our main hypothesis .
If the conductivities of such mixtures are calculated according to the so-called " " law of mixtures they all come out about too high ; but calculated as below they come within of the correct values .
The residual ce , both with the densities and the conductivities , may be due to the slight difference in ionisation in normal solutions which we attribute mainly to a difference in autolytic ionisation .
According to our hypothesis we have , for the conductivity of the solutions of KC1 and respectively , ' where stands the sum of the reciprocals of the radions in each case .
If we mix a volume of the KC1 solution with a volume of the solution , and there is no contraction , we have for the concentrations of KC1 and respectively , in the mixture Hence , if be the viscosity of the mixture , we get for the portions of the conductivity of the mixture due to the KC1 and respectively , assuming the coefficients of ionisation to be equal , and or and THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
and the sum of these portions gives the combined conductivity This , then , should be the true formula for the conductivity of mixtures of equal ionisation .
The formula for calculating the densities of the mixture is , of course , simply In the following table are set out the observed values of the viscosities , densities , and conductivities , together with the calculated values for the densities and conductivities according to the above mixture formulae .
The calculated values for the viscosities of the mixtures are dealt with elsewhere .
TABLE X.\mdash ; Mixtures of Normal KC1 and Solutions .
( e ) Hittorf Numbers Function of the Radions.\mdash ; From equations ( 10 ) and ( 6 ) we see that , according to our theory , and therefore , since , we have Hence the reciprocal of the Hittorf number should be a linear function of the ratio of the radions , and the radions for the ions of a given solute are inyersely proportional to the migration numbers of the ions .
* The most useful way of testing this relation for KC1 and , since it gives us certain important results and verifies the result predicted by our theory in a form * This simple relation leads to a different method of arriving at the radions from that which is followed in this paper , but it cannot be developed here .
VOL. CCVI .
\mdash ; A. independent of any hypothesis , is to take the relation in the form shown in equation ( 12 ) , viz. , ( 12 ) , which gives as linear function of , and to set out the values of as ordinates npon the values of as abscissae .
This is done in Diagrams II .
and III .
In the following tables are set out the values of the Hittorf numbers TABLE XI.\mdash ; Hittorf Numbers for , calculated from the Radions .
TABLE XII.\mdash ; Hittorf Numbers for , calculated from the Radions .
according to KOHLRAUSCH ( ' Leitvermogen der Elektrolyte , ' Ed. 1898 , p. 201 ) .
There are also set out the values of calculated from the values of given by KOHLRAUSCH , and the yalues of , taking , for and , for .
There are also given the values of the Hittorf numbers calculated from the expressions for KC1 . . . . .
and for . . . . ?
It will be seen that the agreement between observed and calculated values is very close .
This linear relation is quite independent of the theory , bub predicted by it , and therefore affords strong support to our hypothesis .
THE PHYSICAL PROPERTIES OF AQUEOUS SSOLUTION .
Diagram II .
Diagram III .
Ml W. : IONIC SIZE IN RELATION TO value of these equations considered merely as empirical extrapolation formulae , apart from the theory on which they are based , is clear .
Putting , they us for the values of the Hittorf numbers at infinite dilution the figures and respectively , and these are undoubtedly the numbers which must be obtained by extrapolation from the data given by KOHLRAUSCH .
But in KOHLRAUSCH'S 1900 paper ( loc. cit. ) he states , without giving further data , that he takes the value of the ration number for KC1 as This paper is still the standard paper for the most accurate values of the conductivities of binary electrolytes with monoyalent ions .
The great accuracy of the values of is shown by their correspondence both with KOHLRAUSCH'S empirical formula with the formula given in this paper\mdash ; the modified Van Hoff law with the corrected coefficient of ionisation .
KOHLRAUSCH gives the values of the separate mobilities at infinite dilution as , Na , , Cl , and the values of at infinite dilution as The Hittorf numbers which correspond to these mobilities should be and Both of these diverge seriously from the actual values obtained by extrapolation from the data formerly given by KOHLRAUSCH , especially the number for , which is as against .
It might be well to reconsider the values of the mobilities of the separate ions , in the light of the above given extrapolation formulae , unless it turns out that KOHLRAUSCH had more accurate but unpublished data in possession in 1900 , since these numbers are of such fundamental imporin the theory of conductivities .
( f ) Deduction of Numbers and Values of Radions Ionic \mdash ; It remains now to deduce the hydration numbers for our present purpose .
Comparing the expressions for above given with equation ( 12 ) above set out , the KC1 equation gives us for the Cl ion the value , and the equation gives us for the ion the value .
This is on the hypothesis that the values of the numbers at infinite dilution are and respectively .
We shall assume , however , that KOHLRAUSCH has data at his disposal which justify his taking the figure for KC1 at instead of , which makes the velocities ions equal , and we shall provisionally fix the hydratio1l number for Cl at the .
We then deduce the values for the and Na ions from the equation THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
and thus get for the values for the and Na ions , The complete series of hydration numbers is then as With the aid of the numbers for the separate ions the radions are calculated from the expression taking the values for given by KOHLRAUSCH in his paper and above set out .
The values so obtained are set out in the following tables:\mdash ; TABLE XIII.\mdash ; Values of Radions and Ionic Volumes in KC1 Solutions .
TABLE XIV.\mdash ; Values of Radions and Ionic Volumes in Solutions .
In the above tables the values of the radions are given in ionic units as before defined .
Expressed the same unit the volume of an ion would be .
But we shall save trouble , with a gain of convenience , if we take the volume of the ions as .
Our unit of ionic volume is therefore the volume of an ion whose radius measured in ionic units is .
The volumes of the ions so calculated are also set * These numbers can only be regarded as approximately fixed .
The existenoe of the discrepancies above indicated in the values of the HITTORF numbers introduces considerable uncertainty .
This , however , is not of much consequence in the consideration of dilute tions , since the magnitude of the radion is primarily fixed by the mobility at infinite dilution , the }tion number affects only the rate of change of magnitude with dilution , which is small .
out in the tables , together with the total volume of the pair of ions , which is indicated by the letter , and is the sum of the volumes of the two ions .
The above are the figures which will most frequently be required .
Where the values for other concentrations are wanted they can be calculated in the manner above indicated .
It will be noticed that the volume of the sodium ion greatly exceeds that of either the or Cl ion , indicating that the sodium ion is combined with about four times as much water as either of the other two ions .
PART IONIC SIZE 1N RELATION To DENSITY .
Solution Volume Linear Function of Ionic Volume.\mdash ; As explained in the former paper , the solution volume derived from density observations is a measure of the amount of contraction which takes place on solution , and therefore a measure of the amount of water entering into combination with the solute .
The ionic volumes at different dilutions should be nearly proportional to the amounts of water in combination with the ions , if the volume of water is large compared with the volume of the ionic nucleus .
Hence , if our theory is correct , we ought to find the solution volume approximately a linear function of the ionic volume of the pairs of ions in dilute solutions .
In the following tables are set out the solution volumes for KC1 and solutions from to 6 per cent. , together with the ionic volumes taken from Tables XIII .
and XIV .
The solution volumes are calculated from the densities by means of the formula , THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
where is the density of the solution at C. and the density of water .
The KC1 densities used are given in the former paper .
The densities used are those tabulated in Table I. In diagrams IV .
and V. the solution volumes are plotted as ordinates upon the IO II I3 I4 I5 Diagram V. ionic volumes as abscissae .
They give the following linear relations , from which the calculated values of V*given in the tables are derived* :\mdash ; For KC1 for . . . . .
TABLE XV.\mdash ; Ionic and Solution Volumes for P. P. P. P. P. P. P. P. P. P. P. P. ( from density ) .
0 .
377 0369 0367 0 363 calculated .
Difference .
*In the former paper the hydration numbers and for and Cl were taken as equal .
The revised figures for KC1 are therefore here given .
.\mdash ; Ionic and Solution Volumes for ( b ) Density Function of Radions.\mdash ; As in the former paper , we may now proceed to deduce a rational density formula for KC1 and from the expressions for the solution volumes .
The densities are given by the expression .
As V. has been expressed in each case as a function of the radions , we now have the densities expressed simply as a function of the radions .
The agreement between observed values and values calculated from the radions is shown in the following tables:\mdash ; TABLE XVII.\mdash ; Density of , calculated from Radions .
P. TABLE XVIII.\mdash ; Density of , calculated from Radions .
In both cases the formulae hold with great accuracy .
THE PHYSICAL PROPERTIES OF SOLUTION The above results are calculated as if the solution were fully ionised , the ionic volume for a pair of ions bein taken as the sum of the volumes of the ions .
Now , ionisation could only be complete for very dilute solutions , but we find that our formulae hold without sensible loss of accuracy nearly up to normal solutions for which the coefficient of ionisation is less than .
This is a matter of some practical importance hereafter , as it is an indication that we can take the ionic volume of the un-ionised molecules as bein approximately equal to the sum of the volumes of pairs of ions with radii given by the f'ormulae obtained for the separate ions .
This means that if we consider apair of ions just after ionisation to be represented by the figure , we may consider apair of lons before lonlsation to be represented by .
Hence , while a consideration of conductivities and mobilities only gave us a clue to the size of the ions themselves , the density formula indicates that in solutions of moderate dilution the volumes of the un-ionised molecules may be taken to follow approximately the same law .
PART V.\mdash ; IONIC SIZHj 1N RELATION To VISCOSITY .
( a ) Viscosity Function of the Radions.\mdash ; We discuss further on the general relation between the viscosity of a fluid and the sizes of the molecules of which iagram V it is composed .
For the moment , it will suffice to say that it seems , a probable that the change of viscosity caused by the introduction of foreign molecules VOL. CCVI.\mdash ; A. 1.30 MR. W. R. BOUSFIELD : IONIC IN RELATION TO into water would depend upon the number and size of the molecules so introduced .
According to the theory postulated , an ion is a compound molecule consisting of the ionic nucleus with its of water molecules , as are also the un-ionised molecules of the solute .
Hence , if the theory be true , we should expect to find a close relation between the viscosity of the solution and the number and size of these aggregates .
The number of molecules depends upon the concentration , and the.size upon the radions ; the product should therefore be a measure of the joint effect of the introduced ions upon the viscosity in dilute solutions .
Diagram VII .
In the following tables are set out the concentrations , the sums of the radions , the products and the observed and calculated viscosities for KC1 and In Diagrams VI .
and VII .
the viscosities are set out as ordinates upon the products as abscissae .
It will be seen that the viscosity is approximately a linear function of , the expressions being for KC1 . . .
( 14 ) , for . . .
( 15 ) .
THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
TABLE XIX .
\mdash ; Viscosity of KC1 Solutions , calculated from KC1 Radions .
TABLE \mdash ; Viscosity of Solutions , calculated from Radions .
Moreover , if our reasoning is correct , the viscosities of mixtures of normal solutions of KC1 and ought to be given by the formula ( derived from the two preceding ) ( 16 ) , where represents the fraction of the volume of the mixture which consists of normal , and the fraction which consists of normal in this case is unity , since we are dealing with mixtures of normal solutions , and taking the sums of the radions at normal dilution from the preceding tables , the expression becomes .
In the following table the observed values of the viscosity of mixtures of normal and solutions are taken from Table VI .
ante , and the calculated values are those given by the above equation .
* The observed values of for KC1 from to 6 per cent. are taken from the former paper .
The values for the normal solutions of KC1 and are taken from Table .
Being taken from a different series of observations , they do not range well with the percentage series , but the discrepancy , for our present purpose , is not material .
TABLE XXI.\mdash ; Viscosity of Mixtures of Normal KC1 and Solutions , calculated from the Radions .
The agreement between observed and calculated values Tables XIX .
, XX .
, and XXI .
, is sufficient to show that the chief element which determines the viscosity of the solutions is the product .
But the expressions ( 14 ) , ( 15 ) , and ( 16 ) , are not put forward as true viscosity formulae for KC1 and .
They are only intended to show that the viscosity of the solutions is dependent on the size of the .
The results tend strongly to confirm the fundamental hypothesis as to the character of the magnitudes which we have called " " radions and which we derived from the conductivities by the application of the Stokes theorem and the Van Hoff law .
It is surprising that the agreement with the observed values in Tables XIX .
and .
should be so close when we consider that at normal concentration the ionisation is less than .
This may be explained on our theory in the following way .
As indicated by the agreement upon the density formulae ( see Tables XVII .
and XVIIL ) , there appears to be no in tloe law as between ionised un-ionised molecules .
Thus , if represents the condition of things when the molecule is just sufficiently hydrated for ionisation , and ( b ) the condition of things shortly before ionisation , the same functions approximately express the ionic radii before and after ionisation .
A little consideration will show that the expression takes the radion of the un-ionised molecule as .
For if be the ionisation , this would give us for the separate ions the products and , and for the un-ionised molecule the product .
The sum of the three products is simply .
This seems to explain why the linear relation , which might be expected to obtain between and for dilute solutions , does in fact approximately persist up to normal solutions .
( b ) References to Previous Theories.\mdash ; The results obtained in the last section suggest considerations which may tend to a great simplification of our ideas in reference to the internal friction of fluids .
Hitherto , very diverse theories have been advanced in order to account for the apparent anomalies in the viscosity of liquids , and various general formulae have been proposed in order to express the experimental THE PHySICAL PROPERTIES OF AQUEOUS SOLUTIONS .
facts .
proposed a formula of the form ( ' Zeit .
Phys. Chem 1 , 285 , 1887 ) to express the viscosity of a solution in terms of the concentration .
For a mixture he proposed .
This formula in the case of , where the constants the two ions could be taken as approximately equal , reduced to the form .
In this form EULER Zeit .
Phys. Chem further investigated the matter .
Neither of these formulae turned out to be of general validity .
GRUNEISEN* ( ' Wiss .
Abh .
der Phys.-Tech .
Reichsanstalt , ' 4 , 239 , 1905 ) has elaborated a very accurate formula which takes the form where is the coefficient of ionisation .
GRUNEISEN has given the constants and for a large number of electrolytes , and the formula brings out the important fact that the equivalent viscosity increment for electrolytes has minimum value at concentrations varying from normal to normal .
But GRUNEISEN'S results stop short of furnishing a theoretical explanation of the facts .
He observes ( loc. cit. , p. ' ' That the minimum found with electrolytes should result from experimental errors , I regard as impossible .
The only doubt is whether it is peculiar to electrolytes .
I held this idea at first , but further experiments are certainly desirable .
This suggests the possibility that the phenomenon may be accounted for by the dissociation of the molecules .
On this supposition , the increase of the equivalent viscosity increment at greater dilution would be most simply explained by assuming that the dissociation of the molecule always increases the riction of the solution Other observers who have theorised on supposed viscosity anomalies have tendered widely differing explanations .
ARRHENIUS Zeit .
Phys. Chem , referring to the lowering of viscosity , or so-called " " negative friction which results from the solution of certain substances , such as , suggests that the lo of the friction is result of electrolytic dissociation .
EULER ( ' Zeit . .
Phys. Chem 25 , 541 , 1898 ) suggests that the change of viscosity in a solution consists of two components , first , an increase which is in inverse proportion to the mobilities of the ions , and , secondly , a component depending upon the compression ( electrostriction ) of the water through the electric charge of the ions .
WAGNER ( ' Zeit .
Phys. Chem 46 , 871 , 1903 ) controverts both these views , and adumbrates the view ( based upon mixtures of alcohol , nitro-toluol , and other organic compounds ) that the viscosity of a mixture depends upon the viscosity of its components , and that this will account for " " negative friction * GRUNEISEN'S paper came to my notice whilst the present paper was in progress .
His resuIts afford strong confirmatio1l to the hypothesis with which this paper is concerned , and a reference is made to them later .
One matter of great importance has been established in relation to viscosity , viz. , that it is an " " additive\ldquo ; property .
This is due , in the first instance , to the work of REYNER Zeit .
Phys. Chem and WAGNER ( ' Zeit . .
Phys. Chem 5 , 46 , 1890 ) , and finally has been placed on a very firm foundation by ( loc. cit ( c ) A priori Considerations to a Viscosity \mdash ; The results obtained in section appear to point to the simple view that the viscosity of a solution is in the main merely a function of the number and size of its component molecules .
We there saw that the of viscosity of a liquid could be approximately expressed in terms of the size and number of the molecules of the solute introduced , both in the case of simple salts and of mixtures .
But the viscosity of the solvent itself should be capable of similar expression .
According to STOKES ' theorem , the resistance to a very small sphere moving in a viscous medium is proportional to its radius .
A priori this would lead us to the hypothesis that the viscosity of the medium itself would be proportional to , in the case of a simple homogeneous medium consisting of molecules of radius .
Furthermore , this , in conjunction with the result which appears in section , suggests the hypothesis that the viscosity of a heterogeneous medium , consisting of molecules of radii , , in proportions , , could be expressed as In the general case , the different sorts of molecules might have different coefficients of friction , so that the constants would have different values .
But , in the case of aqueous solutions , supposing the ions and molecules to be sufficiently hydrated , we should have , both in the case of water molecules and molecules of the solute , the friction of water on water .
We will COIlfine ourselves to this case and take the constant as the same for the different molecules , so that our expression would be simply This would possibly exclude the case of and OH ions , which are possibly little , it at all , hydrated .
In case of molecules differing much from the spherical form either a special coefficient of friction would have to be assigned to them , or their radion would have to be taken as their largest or some mean dimension .
The numbers might be regarded as expressing either the relative numbers of the various molecular species or their relative amounts reckoned by the total volumes occupied by each species in the solution .
In some 'cases it will make no difference which view we adopt , in others it may make a serious difference .
It is easier to work with relative volumes , and we find that such a volume relation holds for the changes of viscosity of mixtures of normal KC1 and solutions .
We shall therefore adopt this alternative in order to test the hypothesis , bearing in mind , THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
however , that this is merely a tentative selection from the alternatives .
Hence , if , represent the total volumes occupied in the solution by each species of molecules , we propose to represent the viscosity of the solution as In order to simplify matters , we shall always take , so that , will be the fractions of the total volume of the solution occupied by each species of molecules , and the expression becomes ( d ) Extended Conception of tloe Radion.\mdash ; The conception of the radion as the average molecular radius of any set or sets of molecules , basing the average on a volume reckoning , will greatly simplify the hypothesis .
We adopted the term " " radion\ldquo ; as a convenient contraction for the radius of the ion , which we supposed to be an aggregate consisting of the ionic nucleus and water molecules .
The expressions which we obtained for the ionic radius in terms of hydration , which were of the form clearly indicated that the radius so derived must be an radius , since the hydration is a continuously varying magnitude , whilst the actual numbers of molecules must be whole numbers and must vary per saltum .
We may now generalise the conception of the radion and use the term to indicate the average radius of any group of molecules .
Suppose , then , we have a liquid containing a volume of molecules of radius , a volume of molecules of radius , and so on , the radion of this liquid , taking the average on a basis of volumes , would be If we always consider a unit of volume so that , and therefore , are the fractions of the total volume occupied by each set of molecules , then for the radion , or average molecular size , simply It follows also that if , be the radions of different volume fractions of the liquid ( instead of the radii of the ultimate molecules ) the radion of the liquid is still expressed as ( e ) The Viscosity .\mdash ; The general viscosity formula for aqueous solutions in which the coefficients of friction between the molecules may be taken as constant throughout , to which we were led by 1 priori considerations , was The conception of the radion enables us to express this simply as , where is the radion of the whole solution .
Also if the liquid be composed of various composite constituents having radions , for the respective volume fractions , , the viscosity of the separate fractions ( if they could be separated ) would be , , and the viscosity of the whole , if is the radion of the whole , since We have determined the values of the radions for , Na and Cl ions in Part III .
( g ) of this paper .
In extending the conception of the radion to the molecules of the solvent or to other molecu ] , its magnitude in any particular case must be measured in the same ionic units , and by reference to the scale furnished by the ions whose radions we have determined .
Thus we know , upon our theory , the radions of and Cl ions , which are approximately equal , and we find that the introduction of such ions into water makes , speaking roughly , no change in the viscosity of the water .
This tells us at once ( if we neglect the small change in the water radion which is probably produced by the solute ) that the of wclber is of about the same gnitude the radions of the and Cl ions .
As a fact , the introduction of these ions slightly diminishes the viscosity of water and hence the radion of the uncombined water in the solution should be somewhat greater than the radions of and Cl ions .
There is good reason for believing that the introduction of a solute into water somewhat lowers the radion of water , since the process of abstraction of water molecules by the ] breaks up some of the molecular complexes of the uncombined water and therefore lowers the average molecular size of the un-combined water .
The investigation of the amount of this lowering is a complex matter , and we shall neglect it , as it does not seriously affect the general character of our results , though it Volves a slight loss of accuracy .
The fundamental constant of our viscosity equation is calculated from the experimental data in reference to water and KC1 and solutions , but it is not necessary to gi ve the details of the process .
The justification of the value given and also of the theory involved is to be found in the correspondence between observed and calculated values given later .
The value of determined from the data at C. is No closer adjustment of its value is at present attainable .
This value gives the radion for pure water as ( f ) Experimental Very .fication.\mdash ; We now proceed to test the hypothesis by reference to the experimental data before given .
Taking first the simple case of mixtures of KC1 and normal solutions , we have here solutions which do not materially change their state of aggregation upon mixing .
We can therefore take the radions of the solutions before mixing as being identical with their radions when mixed .
The radion of the normal KC1 solution is The radion of the normal solution is PHYSICAL PROPERTIES AQUEOUS SOLUTIONS .
1.37 In this case the value of makes no difference to our results .
The water radiolt is not required , as we can take the radion of the whole of each solution , before mixing , as the basis of our calculation , thus accounting for the total number of molecules both of solvent and solute .
The constant is here introduced as illustrating the procedure in the more complex case .
Care must be taken to distinguish betweenl the radion of a salt solution as a whole , and the radion of the salt in solution .
In the following table are set out observed values of the viscosities for the mixtures , and the values calculated from the equation , where the actions of KC1 and respectively in unit volume of the ABLE XXII.\mdash ; Viscosities of Mixtures of KC1 and Normal Solutions , calculated from their Radions .
Let us now apply the equation to the general case of a salt in solution .
Let be the radion of the uncombined water in the solution ( which may be diflerent from the adion of water itself ) .
Let be the adion of the hydrated solute ( comprising both ions and molecules ) .
Let be the total fraction of unit volulne of the solution which is occupied by the hydrated ions and molecules of the solute .
Then will be the fraction which is occupied by uncombined water .
Then the equation takes the form , or . . . . . . .
( 17 ) .
This may be regarded as the fundamental viscosity equation a simple salt solution .
It may be noted that if we take as the radion for pure water , we have its viscosity , so that for solutions so dilute that the value of the water adion is not sensibly changed by the addition of the solute we may say , which indicates that if the radion of the solute is less than that of water S is negative .
This is in fact the case for , the solutions of which have a less viscosity than water .
Cases of " " negative friction\ldquo ; therefore are readily explicable upon the present hypothesis .
aVOL..tter is simple , but in } ) the above equation in meet with difficulties which are small for very dilute solutions , but increase with the concentration .
Iu considering what we are to take as the radion of the group of un-ionised lnolecules , our difficulties begin .
But we saw that in our rough viscosity mula for KC1 and no serious error up to 6 per cent. was introduced by taking the radion of the un-ionised molecule as the sum of the radii of the united ions .
This would give too large a value , which would yitiate our equations at higher concentrations , but , within the range of concentration with which we are dealing , dues not introduce a serious error .
The calculation of the volumes of the un-ionised molecules presents a further difficulty .
The un-ionised molecules must be combined with less water than the pairs of ionised molecules , but a consideration of the density law showed us that we might without serious error up to 6 per cent. reckon the volumes of the molecules as though they were composed of pairs of spheres , the radii of which followed the same law as the ionic radii .
We shall therefore continue to reckon the volume in this way , though it may give too much volume to the un-ionised molecule , in a proportion which will increase with increasing concentration .
The total fraction of volume of the solution occupied by any set of ions will be roximately proportional to the number and volume of the individual ions , but not exactly .
If be the of the separate ions of an electrolyte ( reckoned in ionic units ) , we reckoned the separate ionic volumes as ving out the multipliel and introducing the multiplier the sake of co veuience .
We also took the volume of either pair of ions or molecule as In to calculate the absolute volumes from the ionic volumes , we must introduce a suitable factor .
The total volume occupied by the hydrated ions and molecules in a litre of solution being approximately proportional to , let us represent the fraction of a unit of volume of the solution which is occupied by the ions and molecules of the solute by will be nearly the same for different molecules , but not quite , since molecules of different volumes do not exactIy displace proportionate numbers of molecules of the solvent .
If be the coefficient of ionisation , the volumes occupied by the ions and molecules espectively will then be .
in unit vulume of suiutiu the volume of the hydrated ions .
and molecules is , , Hence , if is the radiou of water , the gen eral equation ?
becomes . .
, PHySICAL OF AQUEOUS SOLUTIONS .
1.3 which reduces to , where is of course ( vide equation ante ) the radion of the solute , and the equation , when written in the form , is seen to be identical with the general equation ( 17 ) before given .
In the following tables are set out the values of and , calculated from the above expressions , together with the necessary data for the calculation .
The ionic data are taken from Tables XIII .
and XIV .
The values of are obtained by setting out the corrected values of ( given in Table IX .
) on a large-scale curve and taking off the values for the different values of .
The observed values of the viscosities are the same as in Tables XIX .
and .
The values of used for the calculation were for KC1 and for .
These are mean values calculated from the separate sets of figures for the two salts .
The value of , the radion of the uncombined water , is taken as in both cases .
The value of was worked out from these two sets of figures and the mean value was obtained , as before mentioned .
Considering the various slight inaccuracies involved in the approximations above detailed , the correspondence between observed and calculated values appears to be sufficiently close to support the general character of the hypothesis as to the relation between the radions of the various species of molecules and the viscosity of the solution .
The two sets of viscosities are calculated from the same formula , ) practically same constants , the only difference in the constants employed that the factor is taken as for KC1 and for It must be borne in mind that the purpose was not to construct an accurate empirical formula , which is fairly easy where plenty of constants are available .
But in the cases examined the simple formula is found to be adequate to express , with a near approach to accuracy , the viscosities of the solutions up to normal strength in the case of KC1 and 5 per cent. in the case of , showing that , with such aqueous solutions , the viscosity depends upon the size of all the molecules present .
The departure of the calculated ] from the observed for the solutions at about normal strength may be partly due to our approximations breaking down before normal strength is reached .
But we are more inclined to attribute this departure to a cause which was considered in Part III .
( c ) of this paper .
With an autolytic solute , it is probable that there is some association of the olecules of the solute in more concentrated solutions .
In the case of with its comparatively small ions and molecules , this does not appear .
But in the case of , with its larger hydrated molecules , the radion of the solute might be notably increased by the THF , SOLTTTIONS .
presence of a small fraction of the solute in the form of associated molecules .
in our view , probably accounts for the observed viscosity values for in the neighbourhood of 6 per cent. coming out notably higher than the ] ated values .
It would be going too far from the purpose of this paper to follow up the considerations which are suggested by the view of the relation of viscosity to molecular size which is here presented .
It must suffice to point out that to all the various speculative questions as to the meaning of " " negative friction as to the effect of dissociation in increasing or decreasing friction , and so on , this theory has but one reply\mdash ; the change in viscosity in aqueous solutions under varying conditions is mainly dependent upon the change in the average molecular size which takes place , that is , to the change in the radions of the components of the solution .
( g ) .\mdash ; The important paper by which came to hand luring the progress of this paper , to which reference has already been made , tends strongly to confirm the propositions here advanced .
gives viscosity measurements for some nineteen solutions of electrolytes , and finds that in all cases , as dilution increases , the equivalent viscosity increment comes to a minimum , and then increases with further dilution .
This is in accord with our results , according to which , as dilution takes place , ionisation at first diminish .
the viscosity by breaking up the molecules and afterwards increases it , owing to the increase of the ionic radii with further dilution .
Moreover , GRUNEISEN has dissected the equivalent viscosity increments due to the separate ions at infinite dilution , and has found values of the equivalent viscosity increments for hypothetical " " normal solutions of the separate ions For monovalent ions he finds that these viscosities may be approximately expressed by the equation where is the viscosity of a " " normal solution of ions\ldquo ; and is the mobility of such ions .
The viscosity of water is reckoned as unity .
This result is not exactly comparable with the present results , but an approximate comparison may be made .
If be the radion of the uncombined water , the radion of a set of ions of mobility the fraction of unit volume of the solution occupied by 1 gram-molecule of the ions , we should have , according to .
theory , for the viscosity of the hypothetical " " normal solution of ions If be the radion of pure water , we should have ) , and therefore , for the change in viscosity caused by the introduction of 1 gram-molecule of the ions , we should have ( remembering that according to our theory ) expression , for the purpose of comparison , may be written But according to our view , the value of differs for each ion , and hence GRUNEISEN'S expression , in which and are given constant values for all monovalent ions , could only approximately represent the general result , and in fact it does not profess to do more .
But the important point is that empirical result , which brings out the equivalent viscosity increment of a " " normal solution of ions\ldquo ; as approximately a linear function of the reciprocal of the mobility , is in complete accord with and so far confirms the general theory of ionic sizes herein set forth.$ PART IONIC SIZE 1N RELATION To FREEZING-P01NT DEPRESSION .
General Considerations.\mdash ; As is well known , the so-called " " molecular\ldquo ; freezing- point depression , which is normally nearly constant for dilute solutions of nonelectrolytes , varies with the concentration in the case of dilute solutions of electrolytes , and tends at great dilution to a value which is for binary electrolytes about double the value for a dilute solution of non-electrolytes .
This may partly be attributed to the ionisation of the electrolyte , upon the theory that the ionisation of a molecule of a binary electrolyte breaks up one molecule into two ions , each of which ions produces the effect of a molecule in the depression of the freezing-point .
Hence , if obseryed freezing-point depression , -molecules solute 1000 grammes of solvent , ionisation coefficient , the number of gram-molecules per 1000 grammes of solvent is that is to say , .
Hence , upon this theory , the true molecular freezing-point depression is , instead of , the ordinary so-called molecular depression .
In order to distinguish the fraction it will be convenient to refer to it as the " " effective molecular freezing-point depression or more shortly by the initials E.M.D. , since upon the ionisation theory is proportional to the actual or e.ffective number of molecules present .
When we have evaluated the .
for any electrolyte , we still find it not constant but variable with the.concentration , and it is probable that this variation is intimately connected with the varying ionic volume of the solute , which represents , upon our hypothesis , the varying amount of water in combination with the ions .
relation for divalent ions is .
This appears to indicate that divnlent ions a different coefficient of friction from ions , possibly because they are not approximately sphericn ] .
THE PHySICAL PROPERTIES 143 ' Now , in order to make a just comparison , we ought to consider the value the E.M.D. in relation to the ionic volumes in the neighbourhood of the freezing-point .
But the data at or near C. necessary for making the comparison ately do not exist .
We have for KC1 and obtained ionic volumes and the ionisation coefficients at C. , and it is probable that the of these values will not be sufficiently different at C. to vitiate our comparison , though the values themselves will be sensibly different .
We will therefore make a comparison on this basis .
( b ) -point Depres.a Linear Function of Ionic Volume.\mdash ; JAHN ( ' Zeit .
Phys. Chem 1904 , 50 , p. 136 ) gives a series of values of the molecular freezin depression for KC1 and which we will use as the basis of our comparison .
All his observations are taken in pairs , on solutions of nearly equal concentration .
In most cases we have taken the mean of his duplicate observations , but in a few cases the separate results are taken , in order to in dicate the order of his maximum errol .
S of observation .
In the following tables are set out for KC1 and the following data : , the number of gram-molecules of the solute per 100 grammes of watel\ldquo ; , the hydration , i.e. , the number of gram-molecules of water per gram-molecule of solute , , the adions , of the cube is calculated , as described in Part III .
, , the ionic volume , which in units is equal to the sum of the cubes of the adions , the molecular depression of the freezing-point as give by JAHN , , the true ionisation coefficient at C. , * , the etfective molecular depression of the freezing-point .
TABLE XXV .
TABLE XXVI .
In the last columns are given the calculated values of and the ences f the observed values .
In diagrams VIII .
and IX .
are set out the values of the effective molecular Diagram VIII .
depression as ordinates , upon the values of as abscissae .
stands for the value of the ionic volume at infinite dilution ( see Tables XIII .
and XIV The values of are for KC1 . . . . .
for . . . . .
It will be seen at a glance that we have a linear relation between the E.M.D. and the ionic volume , and that the limiting of the E.M.D. at injinite ditution is same value been determined for -electrolytes as THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
The deviations from the straight line are within the limits of experimental error .
These limits are indicated by the duplicate values at the dilute ends , which differ by Diagram IX .
in the case of KC1 and by in the case of .
The expressions which give the values of the E.M.D. For KC1 . .
, .
The values of calculated from these expressions , together with the differences from the observed values , are set out in the last columns of the above tables .
The maximum difference in the case of KC1 is , and in the case of , and these are the same as the duplicate experimental differences .
It thus appears that the freezing-point depression , both for electrolytes such as KC1 and , and for non-electrolytes such as sugar in dilute solution , may be expressed by one formula , viz. :\mdash ; since in the case of a non-electrolyte like sugar the ionisation is nil , so that and the molecular volume is nearly constant in dilute solution , * so that * We are justified in saying that the molecular volume of sugar is nearly constant in dilute solution owing to the facts:\mdash ; ( a ) That the solution volume of sugar in dilute solution is nearly constant .
( b ) That the viscosity of dilute sugar solutions is almost a function of the concentration .
VOL. CCVI .
\mdash ; A. Moreover , the value of is nearly the same for both KC1 and , though the ionic volumes are very different .
( c ) Effective Freezing-point Depression Function of Solution Volume.\mdash ; In Part IV .
( a ) it was shown that the solution volume was a linear function of the ionic volume , that is the sum of the cubes of the radions , and it has now been shown that the effective molecular freezing-point depression is also a linear function of the ionic volume .
If this be so , we can dispense with our hypothesis and obtain the effective molecular freezing-point depression as a linear function of the solution volume .
With the object of testing this relation , the density measurements of KC1 and at C. were made .
these have been already derived formulae expressing the solution volume in terms of hydration ( see Part II .
) .
If we denote by the difference between the solution volume at a given concentration , and the solution volume at infinite dilution , the expressions we obtained in Part II .
( c ) may be written for KC1 , for , where having the ] for KC1 and 4 for .
At infinite dilution for , and for By means of these formulae we can calculate the solution volumes corresponding to the concentrations at which JAHN'S freezing-point observations were made , and we find , as our theory led us to anticipate , that the effective molecular freezing-point depression is in fact a linear function of the solution calculated from the densities .
gram X. THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
In the following tables are set out the figures given by JAHN for KC1 and side by side with the solution volumes calculated from the densities .
In Diagram X. are set out the values of the E.M.D. , both for KC1 and , as ordinates upon the values of as abscissae .
The maximum experimental error in the values of the .D .
is , so that within the limits of experimental error we have the E.M.D. a linear function of .
The values of the E.M.D. appear to range a little higher for than for , as was the case when they were set out upon the ionic volumes .
This may represent a real difference , as there is reason to suppose that the density of the water , in combination with the ions and molecules , is slightly different in the two cases .
In the comparison with ionic volumes we gave two different expressions , with constants of slightly different value .
In the present case we will take one single formula to represent the E.M.D. for both KC1 and The calculated values of are set out in the tables , and the differences from the observed values given in the last columns .
TABLE XXVIL Difference .
TABLE XXVIII .
If , as in the comparison with ionic volumes , we take two slightly different constants for KC1 and , we obtain differences between observed and calculated values which are substantially identical with those given in Tables XXV .
and XXVI .
Since , in the case of non-electrolytes such as sugar , we have in dilute solutions the solution volume practically constant and the ionisation nil , we arrive again at an expression which is valid both for electrolytes such as KC1 and , and nonelectrolytes such as sugar , viz. : This result is , of course , quite independent of the hypothesis which it is the purpose of this paper to test , and gives us a ready means of extrapolation for arriving at the true molecular freezing-point depression at infinite dilution based on simultaneous density observations .
Whether expressed in terms of ionic volumes ( according to our hypothesis ) or in terms of solution the meaning of the relation seems to be clear .
Both and are proportional to the decrease in the amount of water in combination with a given weight ( say a gramme equivalent ) of the solvent as concentration increases .
Hence we get the result that in dilute solutions the increment of the effective molecular depression of the freezing-point above as concentration increases is proportional to the decrement of the of water in combination with -molecule of the solute .
To discuss the physical meaning of this relation would carry us too far from the main lines of the present paper .
( d ) Correction of Freezing-point Depression by Reference to Ionic Volumes.\mdash ; In Part V. ( f ) our experimental results were shown to lead to the view that the viscosity of an aqueous solution was proportional to the average molecular size of all the molecules .
We were able also to calculate approximately from the viscosity data the actual volume occupied by a gramme equivalent of the solute at a given concentration .
If be the ionic volume at a given concentration , and the volume in litres of a gramme equivalent of ions be taken as , it was found in the case of , from the viscosity data , that .
Hence the volume of a gramme equivalent of KC1 in solution is , reckoned in cubic centimetres .
To find the volume of the KC1 itself , we have the density of solid fused KC1 which is given by QUINCKE as , which gives for the volume of 1 gramme of liquid KC1 cub. centims. , allowing 3 per cent. for expansion from the solid to the liquid state .
This gives the volume of a gramme equivalent of liquid KC1 as cub. centims. , and therefore the volume of water combined with 1 gramme equivalent of KC1 is cub. centims. Hence , if we have a solution of KC1 containing gram-molecules dissolved in 1000 grammes water , we have approximately for the amount of free water .
PHYSTCAL OF AQUEOUS SOLUTIONS .
Let us now calculate the effective molecular freezing-point depression , not upon the total amount of water present but upon the amount of free water obtained from the above expression .
The expression for the E.M.D. calculated in this way is ' where is the number of gram-molecules of the solute in 1000 grammes of free water .
We have then and In the following table are set out the values of resulting from JAHN'S data ( taken from the column " " calculated\ldquo ; in Table XXV together with the other data necessary for the calculation of , the effective molecular depression of the freezingpoint calculated on the amount of free water .
It will be seen that we obtain value for , up to about twice decinormal solutions , the mean value of the constant being TABLE XXVIIIA .
from m We may recapitulate the method by which the above figures are obtaiJJed .
The ionic volumes measured in arbitrary units having only relative values were deduced from the corrected conductivity data .
The factor which was required to give us the absolute volume for a gramme equivalent of KC1 in solution , with its combined water , was deduced from the viscosity data .
The volume of free water so deduced , when applied to calculate the effective molecular freezing-point depression , gives us for this constant value , which is in correspondence with the value obtained for non-electrolytes .
We thus obtain a striking confirmation of the validity of the hypothesis and of the method pursued , and at the same time illustrate the dependence of the various physical properties considered upon the varying size of the hydrated ions caused by the varying amount of water with which they combine .
* * The corresponding values for solutions give the constant as for high dilutions , but for lower dilutions the value is too low , indicating over-correction .
This may indicate that either the ionic volumes *duced for , or the value for , may be slightly too high .
The larger ionic volume for the Na ion renders any error much more effective in disturbing the constant .
PART IONIC SIZE 1N RELATION To REFRACTIVE POWER .
The refractive power of a medium is known to be closely related to the state of aggregation of its components .
According to the hypothesis which we are discussing , the combination of water with the ions and molecules of a solute changes the state of aggregation of the fraction of the water which is so combined , and also , to some extent , of the uncombined water .
Hence we should expect to find a close relation between ionic volume and refractive power .
If refractive index of a solution at C. , refractive index of water at same temperature , concentration ( gramme equivalents per litre of solution ) , the simplest measure of the refractive power is , say , which we may refer to shortly as the equivalent refractive power of the solute .
A recent paper by DJNKHAUSER Sitz.-Ber .
, Wien .
Akad .
Wiss July , 1905 , p. 1001 ) gives the values of this equivalent refractive power for solutions of KC1 and .
The values for KC1 appear to be a little rough , but the values for seem to be particularly accurate .
* In the following tables are given the concentrations , the values of and of the radions and ionic volumes calculated therefrom , together with the values of the equivalent refractive power given by DINKHAUSER .
In Diagrams XI .
and XII .
are set out the values of the equivalent refractive power as ordinates upon the values of the ionic volumes as abscissae .
It will be seen that a linear relation appears , which may be expressed by the equations for KC1 for The values calculated from these expressions and the differences from observed vaIues are also given in the tables .
If be the value of at infinite dilution ( see Tables XIII .
and XIV and we put , the expressions may be written for KC1 for * For KC1 , DINKHAUSER gives the value for ( p. 1020 ) and the same value for ( p. 1025 ) .
The whole series shows that the lavter value must be inaccurate .
For he gives the value ( p. 1020 ) and for the value .
The former of these values is eiected .
The values for concentrations less than one-fiftieth 1lormal are also rejected , since the large factor by which the observed value of must be multiplied to get the equivalent refractive power greatly magIlifies any error .
These discrepancies are partly due to the fact that DINKHAUSER has combined his own resnIts with those of other observers .
THE PHYSICAL PROPERTIES OF AQUEOUS SOSOLUTIO .
Diagram XI .
Diagram XII .
TABLE XXIX .
TABLE XXX .
ated .
Table XXX .
shows that the relation in the case of down to half normal solutions is particularly accurate .
In Part IV .
( b ) it was shown that the ionic volume might be calculated from the radions of the separate ions with fair accuracy down to nearly normal solutions .
It is obvious , however , that , as the proportion of un-ionised solute becomes greater , it becomes more and more inaccurate to take the ionic volume of a molecule as the sum of the volumes of a pair of ions .
It is probable that the divergence of the calculated values in Table XXX .
for the last two observations tabulated is due to this error in the calculation of the ionic volumes .
Now , as in the case of freezing-point depression , so here , having shown previously that solution volume is a linear function of ionic volume , we deduce that the equivalent refractive power , which is shown above to be a linear function of ionic volume , must therefore also be a linear function of solution volume .
The series of values given by DINKHAUSER for KC1 is not sufficiently good to make this comparison useful .
But the series shows clearly the remarkable fact that the equivalent refractive power is linear function of the solution volume right through the range of dilute and concentrated solutions .
In Table XXXI .
are set out the solution volumes , calculated from the density data THI PHYSICAL PROPERTIES OF SOLUTIONS .
giyen by DINKHAUSER , values of the equivalent refractive power as given by him , together with the values calculated from the linear relation It will be seen that the difference between observed and calculated values nowhere exceeds more than a unit in the last place of decimals , from a concentration of ' one-tenth normal to five times normal .
It may be noted that we have here a clue which may enable us to obtain the ionic volume in concentrated solutions .
In dilute solutions we have shown that the ionic volume , as derived from the radions , is a linear function of the solution volume .
It seems probable that this linear relation persists through concentrated solutions , and that the solution volume may therefore enable us to calculate accurately the ionic volume in concentrated solutions .
But we cannot follow up this clue in the present paper .
TABLE XXXI .
0 1 025 0 .
5 0 .
75 observed .
001035 001023 001008 000994 000982 000946 000922 000898 0 .
00876 calculated .
001034 001022 001007 000995 000983 000947 000923 000899 000876 PART VIII .
\mdash ; CONCLUDING OBSERVATIONS .
Let us now briefly review the main course of the argument , as developed in this and the former paper , in relation to our fundamental hypothesis .
As our starting point we took the Van Hoff dilution law , which we may express by saying that if represents the concentration of the dissociated portion of a solute and the concentration of the undissociated portion , reckoned by means of the * DINKHAUSER gives the values of , where is the difference of the density of the solution from the density of water at the temperature , and is the concentration .
The solution volumes can be calculated from the expression , where is the of a gram-molecule of the solute .
It would be more accurate in concentrated solutions to use the term molecular volume instead of ionic volume , but unfortunately the term molecular volume has been adready misapplied to indicate the product of the solution volume by the weight of a gram-molecule\mdash ; a use which has led to much confusion of ideas .
VOL. CCVI.\mdash ; A. X ordinary value of , there exists a linear relation between and , leading to a relation of the form , where is for various electrolytes nearly but not quite two thirds .
( See Appendix .
) This suggested that if we could find a suitable correction for , which usually taken as , the Van Hoff dilution law would turn out to be an exact relation for dilute solutions of binary electrolytes such as The materials for such a correction were sought in the known fact that the viscosity of the solution produced aberrations in the mobilities of the ions , but viscosity differences alone were inadequate to give an account of such aberrations .
KOHLRAUSCH'S observations on the temperature coefficients of the ions had already led him to the general view that the ions must be considered to be water-coated .
This water colnbinatio necessarily altered the sizes of the ions , and it was considered that the joint elect of changes of viscosity and changes of size might adequately account for changes in the mobility .
reckon these effects quantitatively , STOKES ' theorem as to the motion of a small in a viscous medium was available , and though the actual motion of the ions through an electrolyte under the influence of a potential gradient is probably extremely erratic , it was thought that , nevertheless , the effect of size and viscosity upon the average rectilinear drift under the influence of the electro-motive force might be amenable to exact treatment , just as the average rectilinear drift itself can be accur tely calculated .
Assuming , then , that the aberrations of the mobilities which made the Van Hoff law inexact were due to such causes and could be dealt with in this way , corrections applied to the mobilities , and corrections for were calculated , the nature of which was determined by the Stokes theorem , and the amount of which was determined by the Van Hoff law ( expressed in terms of the hydration instead of the volume V of the solution ) .
The result of this process was to give us the expression for the relation between the radius of an ion and the dilution of the solution .
Up to this point , if one may compare small things with great , the process followed is similar to that of the astronomer who sought to locate the position of a new planet considering the irregularities in the movements of the old ones .
In that case the result could be tested by turning a telescope to the spot indicated by the calculations .
In the present case the result could only be tested by considering how far the othesis as to the changing sizes of the ions owing to the changes in the amount of water combination could be rationally related to the various physical phenomena of solutions , and how the quantitative results were functionally related to existing data .
In the case of each set of phenomena .
it- was- ssar-y to consider a priori how the THE PHYSICAL PROPERTIES OF AQUFOUS SOLUTIONS .
changing size of the ions would be likely to affect it .
A it seemed probable that the Hittorf transference numbers , and the viscosities of the solutions would depend tnerely upon the linear dimensions of the ions , whilst the densities of the solutions and the variations of effective molecular -point depression and refractive power would depend upon the amounts of combined water , and therefore upon the cubes of the ionic dimensions .
These priori considerations have , in fact , turned out to be justified , not merely qualitatively , but with considerable numerical accuracy , having regard to the difficulty of some of the approximations involved .
We are able to express the Hittorf numbers , the densities , the effective molecular freezing-point depressions and the equivalent refractive power within the limits of experimental error , as simple functions of the radions , and to express the viscosities , with a fair approach to accuracy , not merely as a function of the radions of the solute , but also upon the extended conception of the radion as being simply proportional to the radion , or average molecular radius , of the whole solution .
Our hypothesis has also enabled us to discover new relations which are independent of the hypothesis , , the fact that the Hittorf migration numbers are a linear function of , and the fact that the effective molecular freezing-point depression and the equivalent refractive power are linear functions of the solution volume .
It is submitted that the above considerations justify the hypothesis , that the function which we have named the radion , derived as above described , may in fact be taken to be a measure of the actual sizes of the ions .
In any case , the radiol ] turns out to be of fundamental importance in correlating the val.ious phenomena of solution .
APPENDIX .
NOTE 0N THE VAN HOFF LAW .
Reason .
Supposing the Van Hoff Law with Index an Law for certain Electrolytes with Ions .
1 .
A large number of such electrolytes follow the Van Hoff law with close approximation over a considerable range of dilution .
( See BANCROFT , ( Zeit .
Phys. Chem 31 , 191 , 1899 , and for KC1 see SFIELD , 'Zeit .
Phys. Chem , p. 268 , 1905 , where diagram for KC1 is given .
) For KC1 the value of the index is for uncorrected 2 .
In order to test the law with great accuracy , we ought to examine it over al.ange of dilutions so high that the viscosity is practically constant and the increase of ionic size ( if any ) is practically complete .
In such a case we might expect to find the value of practically constant and equal to , if the Van Hoff law holds accurately .
It is impossible to make such a test , owing to the fact that the error introduced by the uncertainty as to the correction for the conductivity of the solvent , multiplied by the MR. W. R. BOUSFIELD : IONIC SIZE IN RELATION TO factor necessary to obtam the equivalent conductivity , entirely swamps the small difference ( if any ) from the exact value of the index which we desire to detect .
Hence the only available means of investigation is to get the true value of at high dilutions , by extrapolation from lesser dilutions where the errors to which reference is made above become almost inappreciable .
For this purpose a range of dilutions is chosen between normal and normal .
As we descend through this range of dilutions ( from higher to lower dilution ) we necessarily get an increasin divergence from the true value of , owing to changes of viscosity and changes of ionic size , but we can extrapolate into the desired region where viscosity and ionic size are nearly constant .
According to the Van Hofl law , in the modified form adopted in this paper , if , we ought to have In the following tables are set out the values of ( uncorrected , that is to say TABLE XXXII .
TABLE XXXIII .
THE PHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS .
, the values of and calculated therefrom and the values of .
In diagram XIII .
are set out the values of as ordinates , upon the values of as abscissae .
Each value of belongs to a value of intermediate between the two values of from which it is reckoned .
As this intermediate value is uncertain , each value of is set out twice , i.e. , to say upon the two values of which lie one on each side of it .
It will be seen from the diagrams that when Diagram XIII .
, at infinite dilution , the extrapolated value of is in each case zero , i.e. , accurately .
This enables us to say that in the region of dilution where change of viscosity and ionic size become negligible the Van Hoff law holds accurately for KC1 and If we were to follow out the same process for ( say ) acetic acid , we should find that the limiting value of was accurately , i. e. , we should get OSTWALD'S law .
3 .
OSTWALD'S law , with its accurate index , can be easily derived from the law of mass action .
There is , therefore , reason to suppose that the Van Hoff law should also be capable of being based upon the law of mass action , which would necessarily give us , for high dilution , some definite and exact value for the index , since the index is the ratio of certain small numbers of molecules , which are necessarily whole numbers .
In the former paper ( see ' Zeit .
Phys. Chem 53 , p. 266 , 1905 ) this matter was discussed , and a hypothesis was set out upon which the law of mass action gives the Van Hoff law with exact index .
This hypothesis may be unsound , but it shows the possibility of basing the Van Hoff law upon the law of mass action when the nature of the equilibrium between the molecules of solvent and solute is better understood .
MORGAN and ( ' Journ. Amer .
Chem. Soc 26 , p. 635 , 1904 ) have arrived at a similar result , based upon the law of mass action , with a wholly different hypothesis as to combination of ions and water molecules .
hypothesis may be accurate , but it strengthens the view that the Hoff can , in some way , be based on the law of mass action , which would give instead of an approximate index .
The above sets of considerations point to the view , that if , the coefficient of ionisation , be corrected for viscosity and ionic size , the Van Hoff law would hold accurately in the lower range of dilution .
It is found that such a correction can be made by the aid of STOKES ' theorem , according to which the velocity of transportation is inversely proportional to the linear dimensions of the ionic complex , which gives the ionic size as a simple function of dilution .
The relation above indicated should be expressible by the equation , when is small .
This equation can be compared with the relation . . . . .
( 1 ) .
Since for considerable dilution is proportional to ] , the relation may be written as . . . . .
( 2 ) .
This is the relation given by KOHLRAUSCH as an empiricnl result ( ' Sitz .
Akad .
Wiss .
Berlin , ' 1900 , p. 1006 ) .
It is shown by him to hold with great accuraoy from decinormal to infinite dilution for a series of nine binary electrolytes with monovalent ions , and is used by him to deduce the values of As equation ( 1 ) is thus established as an adequate empirical ] ation , we may use it by a reverse process to examine the more sxact relation which the diagram indicates when the curve is prolonged .
Equation ( 1 ) may be written Using Napierian logarithms and writing for the modulus of common logarithms , ( 3 ) , and we may write the relation as By differentiation we get ( 4 ) .
Also from ( 3 ) we get by elimin ting the variable and differentiating , whence by substituting in ( 4 ) and writing we obtain the equation . . .
( 5 ) .
This , then , is the more exact equation of the curve which has been indicated as a straight line in the on p. 157 .
The cnrve starts from the origin , and has a horizontal asymptote .
To compare the constants with KOHLRAIISCH'S figures , we note that , or THE PHYSICAL PROPERTIES OF AQU OUS SOLUTIONS .
For KOHLRAUSCH gives the values which gives for the values .l87 , For the tangent of the inclination of the curve near the origin we have for KC1 the value and for the value These agree well with the diagram .
The accurate empirical law deduced by KOHLRAUSCH is , in fact , not sensibly different in the part near the origin from the Van Hoff law for which That is , in the region where , from the present point of view , ionic size and viscosity may be arded as constant , VAN HOFF'S law , there equivalent ctically to const , is to be taken as the starting point for substances of the class here considered .

