•I . ./■ \'

11.

gt; V

; r
llf\' \'
I

■quot;V ■

, ! • ■ •

\\ \'

PROEFSCHRIFT TER VERKRIJGING VAN DEN
GRAAD VAN DOCTOR IN DE WIS- EN NATUUR-
KUNDE AAN DE RIJKS-UNIVERSITEIT TE
UTRECHT OP GEZAG VAN DEN RECTOR MAGNI-
FICUS. Dr. H. Th. OBBINK, HOOGLEERAAR IN DE
FACULTEIT DER GODGELEERDHEID, VOLGENS
HET BESLUIT VAN DEN SENAAT DER UNIVER-
SITEIT TEGEN DE BEDENKINGEN VAN DE FA-
CULTEIT DER WIS- EN NATUURKUNDE TE
VERDEDIGEN OP MAANDAG 1 JULI 1929. DES NA-
MIDDAGS TE 3 UUR

The researches here described were carried out in the labo-
ratory of Professor H. R. Kruyt to whom I shall always feel greatly
indebted. I would like to thank him for having taken so much in-
terest in the work. His extraordinarily stimulating personality
was at all times a source of inspiration and I shall look back upon
the time spent in his laboratory as the most valuable period of
my scientific training.

Further I wish to take the opportunity of saying how much I
appreciated the friendliness shown to me by all those in the van
\'t Hoff laboratory with whom I came in contact, this having con-
tributed very greatly to the pleasantness of my stay in Holland.

I wish to thank Professor Ernst Cohen (Director of the van
\'t Hoff Laboratory) for his unfailung kindness and courtesy to
me while a guest in that laboratory. I also wish to thank Dr. A.
L. T. Moesveld, conservator of the laboratory, for his trouble in
helping me to obtain apparatus and for his kindness on many oc-
casions. Also Dr. H. J. C. Tendeloo and Dr. P. C. van der Willi-
gen, formerly assistants to Professor Kruyt, whose friendship
will be amongst my most pleasant memories of Utrecht.

I particularly wish to thank Professor I. M. Kolthoff whose
suggestions and advice lead to the successful development of
the methods of determining the aromatic nitro compounds and the
p-^enylenediamine used in the solubility experiments.

My thanks are also due to Mr. Emil Hatschek for his kindness
m allowmg me to complete the experiments on gelatin in his
laboratory at the Sir John Cass Technical Institute (London) after
I had returned from Holland.

The word lyotropic was introduced by H. Freundlich ^ to des-
riDe those properties of solutions which quot;have their origin in the
nteraction of the molecules of the solute and the molecules of the
solvent. This interaction of the two kinds of molecules brings
about a change in the internal pressure of the solution and con-
sequently we may describe all those properties such as surface
tension. compressibiUty. solubility, etc. as lyotropic properties,
and the influence that a dissolved substance has upon these pro-
perties as a lyotropic influence.

The effect of the cation and anion are additive. In general it
will be found that these series irn 1,« onbsp;it

As will be mentioned later, secondary influences will sometimes
bring about irregularities, and one or two of the ions will be found
out of place, so that we may find Na gt; Li gt; K or CI gt; NO3 gt; Br,
but in general, where we are dealing with the influence\' of these
ions on a lyotropic property of a true solution or a colloidal solution,
these two series will appear. We will hereafter refer to them as the
lyotropic series. If we find the influence of ions at the extreme ends
of the series almost the same, we may consider the lyotropic influ-
ence to be small; if, on the other hand, their influence differs con-
siderably and we find a quot;spreadingquot; of the series, we may consider
the lyotropic influence to be large.

The classical example of the occurrence of such a series is in
Hofmeister\'s ^ researches on the salting out of a sol of white of egg.
He determined the lower limit of concentration for various neutral
salts at which the sol immediately became turbid, and obtained the
anion series: Citrate gt; Tartrate gt; Sulphate gt; Acetate gt; Chlo-
ride gt; Nitrate gt; Chlorate. Hofmeister showed that a similar
series existed for the influence of salts on other properties of pro-
teins The lyotropic series is therefore often referred to as the
Hofmeister series.

Jacques Loeb in his work on proteins set out to show that pro-
teins are molecularly disperse systems and follow stoichiometric
laws. He claimed that the hydrogen ion concentration was the all-
important factor in determining the condition of these colloidal
systems and denied the existence of the lyotropic series 3. He claimed
to have shown that the Hofmeister series has no real existence if
one takes into consideration the influence of the added salt on the
hydrogen ion concentration, in which case the series could be re-
placed by a simple valency rule, according to which only the
valency and sign of charge of an ion influence the colloidal behavi-
our of a protein, the other properties of the ion having no in-
fluence as long as no constitutional change in the protein mole-
cule occurs.

Without denying the importance of this valency rule and of
taking precautions that the hydrogen ion concentration remains
unchanged, we wiU, in the following chapters, describe experi-
ments m which the lyotropic series is very marked in spite of the
tact that neutral salts which gave only monovalent ions were used
throughout. It will also be seen that the hydrogen ion plays a part

As a matter of fact the lyotropic series can be found more than
once amongst the results of Loeb\'s own researches. Thus in a chap-
ter which he contributed to Bogue\'s quot;Theory and Application of
Colloidal Behaviourquot;! are given the results of experiments on the
influence of salts on the osmotic pressure of a gelatin sol of a con-
stant of 3-8. The results are unfortunately only given graphi-
cally and no numerical values arc obtainable, but from the graph
we can see that the salts arrange themselves unmistakably in the
lyotropic series. Thus we have in order of osmotic pressure:

CI gt; Br gt; NO3 gt; I gt; CNS
^oeb dismisses these facts with the remark that quot;the variations
in the effects of the six salts with monovalent anion are chance
vanationsquot;.

The researches about to be described were undertaken primarily
to throw some light on the lyotropic influences of salts in colloidal
solutions. Kruyt and de Jong a have shown that the stabiUty of
emulsoids depends on two factors, the charge of the particle and the

existence of this hydration, that is to say an equilibrium between
ound and quot;unboundquot; molecules of the solvent. means that
the presence of 10ns. more or less hydrated, in the sol will make it-
self felt by the exertion of lyotropic influences. Before studying
these phenomena however we found it advisable to study the some-
what simpler cases of lyotropic influence in true solution.

\' Bogue; „Theory and Application of Colloidal Behaviourquot;. (McGraw-Hill
1924) P. 65.nbsp;\'

As will be shown, the lyotropic series appears in such phenomena
as the influence of salts on surface tension, viscosity, the maximum
density of water, the solubility of non-electrolytes, the rate of in-
version of cane sugar, etc. We will however first describe some ex-
periments on yet another phenomenon, namely the influences of
salts in lowering the critical solution temperature of phenol-water
mixtures.

I — THE INFLUENCE OF SALTS ON THE CRITICAL SOL-
UTION TEMPERATURE OF PHENOL-WATER MIXTURES

Kahlbaum\'s (quot;zur Analysequot;) phenol was further purified by
distillation. This gave a product which did not turn pink on ex-
posure to the light and which had a melting point of 41® C.

The experiments were made with mixtures containing 36*5 %,
Phenol and 63-5 % water or salt solution. The mixture of 36-5 %
Phenol and 63\'5 % water gives the maximum critical solution
temperature (C. S. T.) for phenol water mixtures.

Glass tubes 1-3 cm. in diameter and 12-0 cm. in length, fitted with
ground glass stoppers were used. About 2 or 3 grams of phenol
was placed in the tube and weighed accurately. The corresponding
quantity of salt solution was then added from a 5 cc. micro-burette
graduated in hundredths of a cubic centimeter so that the mixture
contained 36-5 % Phenol and 63-5 % water. The tubes were then
placed in a copper stand in a four litre beaker filled with water, so
arranged that the tubes were all equidistant from the thermometer
in the centre of the beaker. The thermometer was a standard
instrument graduated in tenths of a degree. The temperature was
then raised sufficiently to bring all the mixtures above their critical
solution temperature, whereupon the tubes were shaken so as to
assure a homogeneous mixture. The flame was now removed and
the temperature allowed to fall. The rate of cooling was found to be
sufficiently constant and was less than 1° C per minute.

As the mixture cools a slight opalescence first appears which
increases until the contents of the tube become opaque, the latter

part of this change taking place most rapidly. It was found very
convenient to take the temperature when the thermometer behind
the tube could just not be seen by looking through the contents of
the tube. In this way results reproducible within 0.2° C. could
easily be obtained.

All determinations were made in duplicate and repeated (i. e.
two tubes were filled and readings taken together as described; the
temperature was then raised and again allowed to fall, so that a
second reading was obtained.

When weighing the phenol it was exposed to the air as little as
possible. Although it is very hygroscopic the error due to this seems
to have been small as seen from the good agreement between du-
plicate determinations. Timmermans also found this error small ^
^^^^ The phenol water mixture gave a C. S. T. of 66-5° C. Determi-
a^ nations by other workers vary from 65° C. to 70° C. (Schreine-
makers and Alexejew 67° C.; Friedländer 66° C.).

sults which were reproducible and comparable to one another than
to aim at absolute values.

The results for a number of strong electrolytes are shown in
Table I. It will be seen that for the chlorides there is for the in-
crease in the C. S. T. a cation series:

The results for HCl and HBr are of interest as showing that very
large changes in the Ph of the mixtures do not bring about any
appreciable change in the raising of the C. S. T. Thus the influence
of O. 1 N HCl is of the same magnitude as 0.1 N KCl, and, further,
the addition 0.001 N HCl to 0.1 N KCl does not appreciablely
change its influence on the C. S. T. This of course would be impos-
sible if the occurrence of lyotropic series depended on the salts

altering the Ph of the solutions as suggested by Loeb The marked
lowering of the C. S. T. in the case of 0.1 NaOH is in agreement with
the findings of other investigators, that substances which are sol-
uble in both components (i. e. phenol as well as water) lower the
critical solution temperature. It is therefore not due to the lyo-
tropic but the chemical properties of the NaOH.

These results are in close agreement with those obtained by
Duckett and Paterson ^ in a paper published just after our results
were completed. They found for the cation series.

^SO^ gt; CI gt; Br gt; I
In a later paper Carrington, Hickson amp; Paterson ^ found that
for both chlorides and sulphates lithium was not in its usual place,
the series being

Na gt; Li gt; K gt; NH4 gt; H.
This is not in accordance with our results. It is interesting to notice
that they found the lithium ion in its usual place (i.e. before sodium)
in some similar experiments on the influence of salts on the C. S. T.
of the system water-butyric acid. (See table 1, page 8).

Before attempting to explain the influence of the salts on the
critical solution temperature, we must first consider the hydration
of the ions involved.

It is generally assumed that many, if not all, ions are more or
less hydrated. This conclusion has been forced upon the various
investigators by a large number of phenomena and many attempts
have been made to obtain numerical values. Generally the results
are calculated so as to give the hydration as so many molecules of
water to the ion. Such results have been obtained, for instance, from
investigations concerning the mobility of different ions.^ Thus if we
calculate from the mobility of the ion the apparent radius which
must be assumed to make the mobility correspond with Stoke\'s

\' Proteins and the Theory of Colloidal Behaviour p. 14 (1922).
» Ducket amp; Paterson, J. Phys. Chem. 29 295 (1925).
«■Carrington, Hickson amp; Paterson, J. Chem. Soc. J27 2544 (1925).

formula, we find this radius to be greater than the quot;atomicquot; radius
of the ion. Assuming this difference to be due to a quot;shellquot; of water
molecules it is possible to obtain figures which represent the num-
ber of water molecules in the quot;shellquot; of each ion. In this way
Remy ^ obtained the following figures:

NH4

lt; 120
lt; 66
17
16
14
13

Remy concluded that\'the hydrogen ion was not hydrated, a
conclusion which may certainly be considered erroneous. Other
values for the hydration of the ions have been deduced by other
workers from the mobilities ^ in similar ways.

There is however, no agreement upon definite values for these
hydration figures. Bjerrum for example finds that the hydrogen
ion bmds eight molecules of water while the potassium ion binds
none, the hydrogen ion being considered to resemble the lithium

It seems to us that one reason for tlie discrepancies in these re-
sults is to be found in the misleading conception of the molecules
of hydration of the different ions being a definite number of mole-
culcs associated with each ion. There is no evidence to justify the-

\' H. Remy, Zeit. f. Phys. Chem. 89 467 (1915) and also Zeit. f. Electrochemie
29 365 (1923).

\' E. W. Washburn, Jahrb. Radioact. Electronik (1908) 493 and (1909) 69-
E. W. Washburn and E. B. Millard. J. Am. Chem. Soc. 37 694 (1915)- Born\'
Zeit. f. Electrochemie 26 401 (1920); Lorenz, Z. f. Electrochemie 20 365 (1923)

idea that there is a definite shell of molecules of hydration imme-
diately surrounding the ion, within which we have quot;boundquot; mole-
cules and outside of which we have unbound molecules.

A molecule of water (or at any rate the simple molecule HgO as
distinct from the more complex nHgO niolecules) is an electrical
dipole ^ with a considerable electrical moment. Consequently
water molecules in the immediate neighbourhood of an ion will
become orientated. If the ion is a cation their more negative ends
will be pointed towards the ion. Thus there will be formed a layer
of more or less firmly bound ions according to whether the ion is
one largely or only slightly hydrated. Around this first layer of
molecules other molecules will tend to orientate themselves, so form-
ing a second layer (in the case of some ions) around which a third
layer will tend to form (if the forces are strong enough) and so on,
the water molecules being arranged round the ion in chains which
may be very roughly compared to a chain of needles suspended
from a magnet. We may conveniently represent a water molecule
as an isosceles triangle the base of which is the more negative end
of the molecule and the apex of which is the positive end. We then
obtain the following scheme:

(It should be noted that the representation of a molecule of water
by a triangle does not presuppose a wedge-like form for the water
molecule. The triangle is only used as being an easy way of repre-
senting diagrammatically the dipolar character of the molecule).

If then we consider a highly hydrated ion such as the lithium
ion, we see that in its immediate neighbourhood there will be a
number of completely orientated, firmly bound, molecules. Outside l^XxAyk:^ gt;
these there will be more molecules, not so firmly bound and perhapsnbsp;• ^ /

not completely orientated. Still further from the ion there will be
less bound, less orientated, molecules, and so on for some distance k^^ fjr
until we reach molecules that are only very slightly orientated
and eventually molecules which do not feel the influence of the ion. ^ ^ ^^

It seems to us that only by some such scheme is it possible to ima-
gine an ion with (say) 50 or 60 molecules of hydration. It is obvious
that they cannot all be directly united to the ion, and yet it is very
difficult to explain the results obtained from mobility experiments
without assuming in some cases such a high apparent hydration.

between bound and unbound molecules. It will be impossible, for /
..example, to speak of the radius of the ^vated ion. For this reason
different methods of calculating the hydration will give quite diffe-
rent results for the number of quot;molecules of hydrationquot;.

We will therefore not concern ourselves with quantitative values
for the hydration, but we may assume that for the cations the re-
lative order of hydration is

SO4 gt; CI gt; Br gt; NO3 gt; I gt; CNS.
These series, as has already been said, are always to be found when
we study the influence of the corresponding neutral salts on phe-
nomena which involve an interaction between the molecules of a
solute (or colloidal particle) and the molecules of a solvent. The
position of the hydrogen ion in this series is more doubtful. Opinions
on this differ so much that it is considered by some the most highly
hydrated of the series and by others not to be hydrated at all.
This later view is almost certainly wrong. As Freundlich ^ has
pointed out, the hydrogen ion, which is simply a positive atomic
nucleus whose diameter is about 10-quot; cm, would have a velocity
far greater than its actual velocity if it was truly not hydrated.
An explanation from a quite different standpoint of its high mo-

bility was put forward by Ghoshs who suggested that H\' and
OH\'ions, being the ions which can originate from the solvent
water, could be continually formed by alternate dissociation and
recombination of the water molecules whereby a transference
of charges would be brought about which would simulate an unusu-
ally large migration velocity.

The experiments with phenol, as will be pointed out in the next
section, point to the hydration of the hydrogen ion being similar to
that of the potassium ion.

Explanation of the Raising of the Critical Solution
Temperature of Phenol-Water Mixtures

The unbroken line in fig. 2 represents the temperature solubility
curve for phenol in water. Below the critical solution temperature
two phases are formed, the composition of these phases approaching
one another more closely as the C.S.T. is approached. Eventually
at the C.S.T. only one phase exists.

Let us now consider what happens if some KCl is dissolved in
the phenol water mixture at a temperature below the C.S.T. The
KCl will distribute itself between the two phases, the larger con-
centration being of course found in the water-rich phase. The ions
of the KCl are, as4we have already explained quot;hydratedquot; and con-
sequently the orientation of the molecules of water will be different
to what they were before the introduction of the salt. The water
molecules which are orientated or bound by the potassium and

chlorine ions will not be so free to take part in holding the phenol
molecules in solution, and consequently the solubility of the phenol
in this phase will be reduced and a certain amount of the phenol will
be quot;salted outquot; into the phenol rich phase. At the same time owing
to the affinity of the KCl for water, water will be drawn from the
phenol-rich phase, the combined results of these two effects being
to make the composition of the two phases more different. Similar
results will also be brought about by the salt dissolved in the phenol-
rich phase, but, owing to very low solubility, this will be of minor
importance.

Consequently, since the composition of the two phases will be
more different at all temperatures, the solubility curve will fall
outside the solubility curve with no salt present, as shown by the
dotted line in figure 2, and the critical solution temperature will be
raised. (That the curve for an added salt does lie outside the curve
for phenol-water has been shown by Timmermans) ^

If instead of KCl we take an equal concentration of LiCl this
quot;salting outquot; effect will be greater, since the lithium ions are more
hydrated than the K ions, the composition of the two phases (at a
temperature below the C.S.T.) will be made more different to one
another than in the case of KCl and consequently the increase of
the C.S.T. will be greater.

In this way we are able to explain all our results, namely, that
the increase of the C.S.T. by a salt is an additive effect of the ions
of the salt, and that the cations increase the C.S.T. in the following
order:

SO4 gt; CI gt; Br gt; NO3 gt; I gt; CNS
that is to say, the more hydrated the ion, the greater will be its
influence in raising the C.S.T.

The effect of sodium hydroxide in lowering the C.S.T. is due to
another cause. Here the electrolyte dissolves in the phenol to form
sodium phenolate. The NaOH having dissolved in the phenol phase
will attract water from the water-rich phase and make the two
phases more alike so that the C.S.T. is lowered. In general, then,

substances which dissolve in both phases tend to lower the C.S.T
This explanation is in accordance with the observation of Martin
Fischer ^ that while the addition of NaOH caused the phenol-rich
phase to swell in volume, the addition of neutral salts caused a
contraction in the following order

KCl and NaCl bringing about the greatest contraction, as we would
expect from their greater hydration.

The results of these experiments can therefore be explained
qualitatively on the assumption of the effect of the added salts
being a simple quot;salting outquot; process, in which the ions of the salt
reduce the solubiUty of the phenol according to their degree of
hydration. In the following chapter we will show that in some
cases the influence of salts in lowering the solubility of a non-elec-
trolyte cannot be wholly explained on this assumption but that
another influence must be postulated. It is therefore possible that
in the case of the phenol-water mixture the second influence will
also be present though not to an extent great enough to effect
the results qualitatively.

\' M. Fischer „Theory of Lyophilic Colloidsquot; 1st. Colloid Symposium (1923)
n 244.nbsp;\'

Although a great amount of solubility determinations of sub-
stances in the presence of salts have been made, there are very few
data which give a complete cation and anion series for any one
substance.

Rothmund ^ carried out solubility determinations of phenyl-
thiourea in the presence of a number of electrolytes. He found the
cation series to be (for nitrates):

Na gt; K gt; Li gt; Rb gt; Cs
where Na decreases the solubility most and Rb and Cs increase
the solubility. The anion series was:

A number of other results are summarised in a table given in
Rothmund\'s book quot;Loslichkeit und Loslickkeitbeeinflussungquot;. The
substances whose solubilities in salt solutions are considered are
however either gases or liquids. In the case of the gases there are
not sufficient data to conclude that the lyotropic series hold for
any particular gas and in the case of the liquid we are dealing with
the rather different case of the separating of two liquid phases. In
the older literature therefore phenylthiourea remains the one sub-
stance whose solubility has been studied in a sufficient number of
electrolytes to show the lyotropic series.

Recently however a paper was published by Linderstrom-Lang 2
which contains some very interesting results. Linderstrom-Lang
determined the solubiUty of hydroquinone and quinone respectively
in various salt solutions. From his results it is apparent that the
influence of the alkali chlorides upon the solubility of hydroquinone

\' Rothmund. Zeit. f. phys. Chem. 33 401 (1900).
\' Comptes-rendus du Lab. Carlsberg xj No. 4 (1924).

vanes greatly according to the particular cation - (thus lithium
chlonde reduces the solubiHty greatly and caesium chloride hardly
at all) while for the solubility of quinone, the difference between the
influences of the various alkali chlorides is much less marked. On
the other hand he found that KI, KBr and KCl lower the solubility
of hydroqmnone to almost the same extent, while the solubility of
quinone is lowered by KCl and actually raised by KBr Here then
we have a marked influence of the cations in thecaseof hydroqui-
none and, it would seem, a marked influence of the anions in the
case of quinone. There is no evidence of a specific influence of either
cation or anion m the other researches in solubility influence al
ready referred to and so we considered these results of particular
interest and worth extending, since with regard to the anions Lin-
derstrom-Lang\'s results are, for our purposes, insufficient.

Linderstrom-Lang tries to connect the marked difference in
the behaviour of quinone and hydroquinone with the fact that
while qumone is an oxidising substance, hydroquinone is a redu-
cing substance. He put forward a theory of quot;chemical polarisationquot;
to explain the influence of the salts on these two substances, sug-
gesting that an Oxidising substance would not only have a tendency
to attract electrons, but also a tendency (for the same reasons)
to attract anions while a reducing substance would have a tendency
to attract cations.nbsp;^

It seemed more probable to us that the real difference between
the two substances lay in the basic and acid character of the qui-
none and hydroquinone respectively, especially as the results ob
tamed by Linderstrom-Lang on the lowering of the solubility of
succinic acid and of boric acid point to these two acids being of the
hydroquinone type.

We therefore tried to find substances of the quinone type and
turned our attention to basic substances. It was necessary to choose
substances which were only slightly ionised, which had not too laree
a solubility and which could be estimated in the presence of the
several salts used. We found it surprisingly difficult to find substances
Which would fulfil these conditions and eventually chose m- and
p-nitro-anihne and p-phenylene diamine. We also used p-nitro-
phenol. Even for these substances it was necessary to develop special
methods of analysis.

About 30 CCS of the salt solution to be used were placed in a Jena
glass bottle of about 45 ccs capacity (A, Fig. 3) fitted with a rubber
stopper together with a suitable quantity of the substance
whose solubility was be determined.The bottle was then shaken
for about twenty hours in a thermostat the temperature of which
did not vary by more than 0-02°C. The determinations were in
all cases made in duplicate — the saturated solution being ob-
tained from an unsaturated and also from a supersaturated solu-

tion Agreement between two such determinations left no possible
doubt that equiUbrium had been reached. After shaking, the flask
was removed from the shaker and placed in an upright position in
the thermostat so that only its neck protruded from the water. The
rubber stoppen was then removed and the glass T piece C attached
to the neck by means of a piece of rubber tubing B. A glass tube D was
passed through C as shown in the diagram and attached to C by
rubber tubing. By blowing air through C the saturated solution
could be blown out of the flask through D. The lower end of D con-
tamed a constriction, below which was placed a piece of cotton wool

through which the saturated solution was filtered. The first few cubic
centimetres which passed through the filter were of course rejected,
and then the remainder blown over into a clean dry flask. The
quantity of the dissolved substance was then determined by a me-
thod of titration suitable for the particular substance.

The determinations of the solubility were carried out at 23-75°C
so that Lindenström-Lang\'s result could be used for comparison.
Owing to the instability of hydroquinone in neutral solution, a
concentration of 0-0In HCl was used throughout. Thus, instead of
determining the solubility in water we determined the solubility
in 0-0In HCl and then compared this to the solubility in a solution
which contained for example, 1-5 mols of KCl and 0-0In HCl. This
was also the procedure followed by Lindenström-Lang.

The concentration of the hydroquinone was determined by adding
a known quantity (an excess) of a standard iodine solution and
titrating the excess with AsjOg in the presence of sodium bicar-
bonate. The actual procedure was as follows:

lOccs. of the saturated solution was made up with distilled water
to 100 ccs. 10 ccs. of this diluted solution was then placed in
a flask with 50 ccs. of 0-1 N iodine solution. 50 ccs. of a 5 %
sodium bicarbonate solution was then added and the excess iodine
titrated with O-ln AsgOa solution.

The solubility of the hydroquinone in 0-0In HCl we found to be
0-6158 mols per litre at 23\'75°C — a determination with a large
excess of the hydroquinone in the solubility bottle and one with a
small excess giving the same result which pointed to a high degree of
purity. This compares to Linderström-Lang\'s figure 0.6180 mols
per litre. Our results are summarised in Table 2 together with some
of the results of Linderström-Lang, (the latter being marked with
an asterisk). Owing to the different solubility for the hydroquinone
found by him we have here only given his results expressed as a
percentage of his figure for the solubility in water (third column).
For LiCl and NaCl he only made determinations at 18°C, but as
the percentage change in solubility does not vary much with
temperature we may use these figures for comparison without in-
troducing an (for our purposes) appreciable error. Also, as he

employed different salt concentrations it was necessary to obtain
the results given in the table by intrapolation.

The quinone was prepared from hydroquinone by oxidation
with potassium dichromate and sulphuric acid. It was twice re-
crystallised from benzene and air dried. As in the case of hydro-
quinone, all the solubilities were made in 0-0In HCl.

The concentration of the saturated solution was determined by
titrating with a solution of sodium thiosulphate.

Solubility in
grams per litre

50-62
60-62

53-98

54-20
47-91
47-91

52-62
52-82

♦ Results obtained from Linderstrom-Lang\'s work,
t On account of the comparatively low solubility of K,SO,. only deter-
minations with 0.5 M KjSO, were carried out. These gave a solubility of
52.72 grams per litre or 78% of that in water. This gives by linear extra-
polation a value of 67 % for ^^ M KjSO,.

To test the purity of the quinone, about 0^25 grams was weighed
off and titrated. This gave 100-2 %. Further, solubihty determina-
tions were mady by (a) shaking 5 grams with 50 ccs. of water and (b)
1-5 grams with a similar quantity of water. This gave 13-98 and
13-97 grams per litre respectively, as shown in the table.

To determine the solubilities of these substances it was neces-
sary to find a method for their titration which would give suffi-
ciently accurate results. In a method described by E. Knecht and

E. Hibbert ^ for the estimation of nitrogroups in aromatic com-
pounds, the substance is boiled with an excess of titanous chloride
and hydrochloric acid and the excess back titrated with a solution
of a ferric salt. This method entails several disadvantages. Thus
B. Diethelm and Foerster ^ from measurements of the reduction
potentials of titanous solutions were able tb conclude that the ti-
tanous ions may discharge hydrogen\'ions with the evolution of
hydrogen. From the result of a careful investigation by C. F. van
Duin ® we may conclude that this decomposition may give rise to
considerable errors when an acid titanous chloride solution is heated,
even when every precaution is taken to exclude air from the solution.
Van Duin was able to obtain concordant results when a correction
was introduced obtained from a blank estimation in which the ti-
tanous chloride was heated for exactly the same time and in exactly
the same manner as in the actual estimation. In our opinion even
this method is not wholly satisfactory since the titanous chloride
during the blank is under other conditions than that during the
actual estimation. He also found it necessary to standardise the ti-
tanous chloride against a nitro compound. Similarly T. Callan and
J. A. Russell Henderson ¹ standardise against p- nitroaniline. In
some cases they obtain low results when working with titanous
chloride which they attribute to chlorination having taken place,
and for this reason they prefer to use titanous sulphate.

For these reasons we set out to devise a method in which the de-
composition of titanous chloride is eliminated and consequently the
obtaining of theoretical results is not dependent on the choice of an
arbitrary standard substance, and also in which the errors due to
chlorination do not arise.

I. M. Kolthoff» has shown that the reduction potential of a
titanous chloride solution is inversely proportional to the hydrogen
ion concentration; that is that the reducing action of titanous chloride
increases on decreasing the acidity of the solution. Hence we thought
it of interest to investigate the estimation of nitrobodies with

titanous chlonde at low hydrogen ion concentration, as we may
expect that the reduction will take place much more rapidly
The addi ion of salts such as sodium bicarbonate or sLum

of Ti(0H)3 and Ti(OH), due to hydrolysis will take place Itquot;
therefore necessary to use a substance such as Rochelle salt or so-
dmm citrate which as well as having a buffer action also tends to
lorm complexes and so keep the titanium in solution

We made some experiments with RdcheUe salt, but, as this dves
a precipitate of potassium bitartrate while titrating with the strone
ly acid titanous chloride solution, we found it more convenient to
use sodium citrate. In the presence of this salt we found that nitro

An approximately O-OSN solution of titanous chloride was made
by boiling 80ccs of the strongly acid 20 o/^ titanous chloride solu-
tion as supplied by Kahlbaum, with lOOccs. of,concentrated hy-
drochlonc acid for one minute and diluting with water to three
litres. The solution was kept in a brown bottle in an atmosphere of
hydrogen. The apparatus (see Fig. 4) is fully described by I. M
Kolthoff and 0. TomiCek ^ A potentiometric method of titration was
employed. The titrations were carried out in a beaker fitted with a
rubber stopper which was pierced with holes for the burette the
platinum electrode and the inlet and outlet of the carbon dioxide
which was passed through before and during the titration A clean
platinum gauze electrode was used, the beaker being connected to
a calomel electrode by means of a glass syphon filled with a satu-
rated solution of potassium chloride. The readings were obtained
by adjusting the pointer on a rolled slide wire until the galvano-
meter showed no deflection. The carbon dioxide was obtained from
a cy inder and was freed from oxygen by passing through two wash
botUes containing a mixture of 5 % titanous chloride and 20 o/
sodium citrate and a third wash bottle containing water.

We at first attempted a direct titration with titanous chloride,
but the reduction took place so slowly in the neighbourhood of the
end point, that, although results could be obtained in this way, from
40 to 50 minutes were necessary for the completion of a single ti-
tration even at 60° C,

We next tried adding a small excess of titanous chloride and back
titrating with iron alum. In this way it was possible to carry out the
whole estimation at room temperature in less than twenty minutes
and obtain accurate results. Titrations at about 50° C were equally
successful but the higher temperature is not necessary and might in
fact be a disadvantage if the substance being titrated is very vola-
tile. Finally the procedure was as follows:

25 ccs. of the solution of nitrobody were placed in the beaker to-
gether with 30 ccs. of a 20 % solution of sodium citrate. A Uttle
solid sodium bicarbonate (less than 0.25g.) was also added, this
giving an evolution of carbon dioxide, so helping to remove the
last traces of dissolved air from the solution. A steady stream of
carbon dioxide (freed from air as already described) was passed

through the contents of the beaker for at least three minutes be-
fore the commencement of the titration. The titanous chloride was
then run m slowly until the violet colour of the solution showed
there was an excess of from 1 to 3 ccs. The violet colour of the
titanous chlonde is so much more intense in the presence of the
citrate than the ordinary hydrochloric acid solution that an ex-
cess of 1 cc. was more than sufficient to give a marked colour while
with 2ccs. in excess the reduction takes place so rapidly even at room
temperature that the back titration with the iron solution may be
begun about two minutes after the titanous chloride has been added
The back titration with the iron solution was then carried out slowly
until the sudden leap in the potential indicating the end point was
reached. This sudden change in potential was sufficiently large to
make the titration accurate to less than 0-1 cc. of iron alum solution
when the latter was only half the strength of the titanous chloride
soluti^on (1. e. 0.025n). The stream of carbon dioxide is of course pas-
sed throughout the entire procedure. When following this procedure

m no case did we meet with errors due to the air not being entirely
removed.nbsp;^

The curves in Fig. 5. give examples of the magnitude of the
jump m potential. The ordinate represents the E.M.F. as measured
against the normal calomel electrode, and the abscissa 1 cc of
reagent in the neighbourhood of the jump.

Standardisation of titanous chloride solution.
The standardisation of the titanous chloride was carried out by

titrating a 0-05n potassium bichromate solution at 50°C as des-
cribed by Kolthoff and TomiCek Recently E. Zintl and A.
Ranch 2 have raised objections to the use of potassium bichromate
as a standardising substance. They claim that the titration is too
slow and that when using the pure iron-free preparation of tita-
nous chloride steady potential readings are not obtained and the
method fails altogether to give a result. We went fully into
this matter and although we worked with both the ordinary pre-
paration containing iron as an impurity and the specially prepared
iron-free titanous chloride supplied by Kahlbaum, in neither case
did we find any difficulty in obtaining accurate results. The titra-
tion is, as Zintl and Rauch say, rather slow, but this is only a dis-
advantage in the first titration where even the approximate place
of the end point is not known; the subsequent titrations may be
carried out comparatively quickly since it is not necessary to take
so many readings. Further we found the addition of a few drops of
CuSOj solution greatly accelerated the reaction while not inter-
fering with the results. On the other hand we found that the stan-
dardisation against a copper salt as recommended by Zintl and Rauch
gave results which were always about 0-5 % to 1 % too high which
is in agreement with the earlier results of Kolthoff and Tomiciek.
Details of these investigations are published elsewhere

The sodium citrate we used was a commercial preparation twice
recrystallised from water; it did not contain iron as an impurity.
In order to know if the 30 ccs. of 20 % sodium citrate we used had
any effect on the results, we standardised our titanous chloride
against iron alum in the presence of this quantity of sodium citrate
as well as in the ordinary acid medium. The results are set out
below :

Thus in the presence of the sodium citrate the titre is 0-2cc. too
high for 30ccs. of 20 o/, sodium citrate. Even after recrystallisation
of the sodium citrate the result was still the same amount too
high. The difference is proportional to the amount of citrate added
as IS shown in the table. We do not know whether this difference is
due to the citrate itself or to some impurity present in it

The following titrations involving smaller quantities of titanous
chlonde show that the error is a constant error of 0.20cc. of titanous
chloride for 30ccs. of sodium citrate and not dependent on the amount
of the titanous chloride used:

none
none
30
30

1.15
1.15
1.00
1.05

In the titration of nitrobodies this correction which is here also
proportional to the amount of sodium citrate must be applied.

From these results we found that it was necessary to correct all
our titration figures by subtracting 0\'2cc from the figure obtained
by subtracting the amount of the back titration from the quantity
of titanous chloride added.

We also thought it advisable to see if the addition of the sodium
bicarbonate could introduce any error:

could not affect results by more than 0-01 cc. and hence the error
is negligible.

The addition of neutral salts was also found to have no influence.
Further details relating to the general appUcation of this method
are published elsewhere

The p-nitroaniUne after two recrystaUisations from alcohol gave
a result about 1 % too high when analysed by the method described
above. After two more recrystaUisations the following results were
obtained.

22-07
22-08
27-77

The ni-nitroaniline was recrystallised three times from\'alcohol
and gave the following results on titration.

23-58
23-55
23-60
23-60

23-56
23-56
23-56
23-56

lOO-O %
100-2 %
100-2 %

The Solubility Determinations were carried out similarly to those
with quinone and hydroquinone, the shaking being carried on for
twenty hours. The temperature of the determination was 25°C.
With the p- and m-nitroanilines the saturated solution was drawn
off by the method already described and lOccs. of the undiluted
solution titrated; with p-nitrophenol lOccs. of the saturated solution
was made up to 100° and 25ccs. of this solution was titrated. Each
titration given in the tables represents a titration on a separate
solubility determination. For each substance solubility determi-
nations in water were carried out with a large and small excess of
the solid substance in the solubility, this being a further check on
the purity of the substances. The salt solutions employed were 0-2
molar (0-1 molar in the case of KjSOJ

Solubility
as a percen-
tage of that
in water

Solubility
In grams
per litre

Water (with 0-15g.)
Water (with 0-15g.)
Water (with 0-60g.)
Water (with 0-60g.)
0-2M KI
0-2M KI
0-2M KBr
0-2M KBr
0-2M KCl
0-2M KCl
0-2M

100-0

106-7
99-2
96-5

93-5
95-3

94-6

With this substance as with the nitrocompounds no sufficiently
accurate method of estimating the concentration of the saturated
solution could be found in the literature and it was necessary to
develop a special method.

p-Pl^nylenediamine forms an insoluble compound with picric
acid. Hence by adding a known quantity (an excess) of picric
acid to a solution of p-phenylenediamine. filtering off the precipi-
tate, and titrating the picric acid remaining in solution, it was pos-
sible to estimate the concentration of p-phenylenediamine solu-
tion. Preliminary experiments showed that the solubility of the
picnc acid coumpound was not high enough to introduce any appre-

ciable error into our experiments. This was done by shaking some
of the picric acid compound over night in picric acid solutions of
similar concentrations to those used in the experiments, and then,
after filtering, ascertaining if there had been any change in the
concentration of the picric acid solution. In all cases no change was
found within the limits of the experimental error of the titration.
An addition of 0-2M KI solution (which might be expected to in-
crease the solubility) also introduced no error.

The determination of the solubilities of the p-phenylenediamine
was therefore carried out as follows: The p-phenylenediamine was
shaken for 20 hours at 25° C in the particular salt solution in a

thermostat, after which the saturated solution was drawn off in the
manner already described for the other substances. lOccs. of the
saturated solution was made up to lOOccs.

To 25ccs. of this diluted solution was added 75ccs. of 0-05026 N
picric acid solution (titrated against Ba(0H)2 solution). This, after
thorough mixing, was allowed to stand until the next morning. The
supernatant liquid was then filtered. SOccs. of the filtrate (the first
portion coming through the filter having been, of course, rejected)
was titrated with Ba(0H)2, using methyl red as an indicator.

The solubility with a large excess of the solid present in the so-
lubility bottle being higher than with a smaU excess (as shown in
table 12), all the determinations were carried out with a small ex-
cess — about 2-5 grams being used for each determination. Here,
as with the other substances, each titration corresponds to the so-
lution obtained from a separate solubility determination.

The picric acids solutions of different concentration were used
in the determinations. In those determinations marked with an
asterisk 7Sees, of picric acid equalled 88-80 ccs. of Ba(0H)2, while
in the other determinations 75ccs. of picric acid equalled 88-50 ccs.

We may add to these six substances for which we have carried
out solubility determinations the results of Linderstrom-Lang for
boric acid and succinic acid. In order to make these results compa-
rable to the other we have obtained the value of the solubilities in

one molar salt solution by Hnear interpolation from the results
which Lmderstrom-Lang carried out at various salt concentrati-
ons, and expressed these results as a percentage of the solubility
m water. Lmderstrom-Lang\'s results were carried out at two slightly
different temperatures, but the percentage figures may be com-
pared. From his results it may be seen that the Hnear interpolation
mtroduces no appreciable error for our purpose, the curve obtained
by plotting the solubility against the salt concentration being
very nearly a straight line.

Finally we may give some of the results of Rothmund and
Biltz on the solubiUty of phenylthiourea in one molar salt solutions
The result for CsNOg is obtained by extrapolation from his results
for a 0.5 molar solution.

The results of the nine substances whose solubilities in salt so-
lutions have been studied are set out in Fig. 6 so that they may
be readily compared. In this diagram the solubility of each sub-
stance m one molar salt solutions is given, these values being
obtamed when necessary from the value for the salt concentration
at which the solubiUties were actually determined by assuming
that the increase or decrease of solubility is directly proportional

to the concentration of the salt. The error so introduced for the four
substances dealt with by Linderstrom-Lang will be quite small
as can be seen from his figures, and even if in the case of the other
substances the error is somewhat greater it will certainly not be
great enough to be of importance for our purpose — namely the
comparison of the relative quot;spreadingquot; of the cation and anion
series. The temperatures at which the solubilities of the various
substances were determined vary from 18°C to 25°C. However, as
has been pointed out the percentage change of solubility for a
given concentration of salt does not vary appreciably with the
temperature, and so here again no appreciable error is introduced.

c,sa

-•TO CI

KCl *CsCI
ILI_»RVÇL

If now we examine the figure, we see that there are four sub-
stances (quinone, m- and p-nitroaniline, and p-phenylenediamine)
in which there is a very marked spreading of the anions and three
substances (hydroquinone, boric acid and succinic acid) for which
there is a marked cation spreading; the two remaining substances,
(phenylthiourea and p-nitrophenol) seem to fall between these two

\' Rothmund, quot;Lôslichkeit and LosHchkeitbeinflussungquot; p. 154 (Leipzig
1907).

.0d,„e and sulphocyanate ions cause the greatest incr^e o S
decrease m the solubility. There are, hoover, a few^ts oufn
order. Tlus for phenylthiourea, UNO, is found .J rKNO.anLn

The influence of the cation or the anion in lowering the solubility
of a non-electrolyte is, as we have seen, generally the greatest when
the ion is most hydrated. For this reason it is frequently assumed
that the influence of the ions is simply one of binding a certain
proportion of the water molecules and so rendering them incapable
of dissolving the substance which is quot;salted outquot;. Debye ^ has more
recently connected the lowering of the solubility of organic sub-
stances by addition of electrolytes with the fact that these organic
substances lower the dielectric constant of water. We have how-
ever shown that cases of increased solubility frequently occur, and
for this reason alone, this theory must be considered inadequate.
Further we have shown that for some substances the anions seem
to have a very great influence compared to that of the cations
while other substances are influenced more by the cations than
the anions. To explain these facts, therefore, it is necessary to assume
that the solubility influence is brought about by more than one
factor and in what follows we will show how such an explanation
is to be found in the orientation of the dipoles of the water mole-
cules.

Let us consider what happens when a substance dissolves in a
liquid. As is well known, for a non-electrolyte to have more than
an extremely low solubility in water, one or more polar groups must
be present in the molecule ^ (witness the very low solubility of
the inert gases, paraffins etc.). When therefore a substance con-
taining such polar groups dissolves it may be considered the forces
of attraction between the water molecules and the molecules of the
solute overcome the forces holding the molecules of the crystal of
the solute together (quot;lattice energyquot;). When the saturation point

\' P. Debye. Physik. Zeit. 26 22 (1925).
\' W. D. Harkins, J. Am. Chem. Soc. 39 354 (1917).

has been reached, these forces may be considered equal and we have
a dynamic equiUbrium. But as the molecules of the solute, although
electrically neutral, contain one or more polar groups, the dipoles
of the water molecules must be orientated around the solute mole-
in a way similar to that we have described in dealing with the
hydration of the ion. Thus if the polar group in question is negative
(e.g. an hydroxyl group) the positive ends of the water molecules
\' .nbsp;will be turned towards it. Further it will be seen that the forces of

: (JUL lily ^ attraction between solute and solvent will be at then-greatest when
\' M jL Jorientation is complete, and that anything that decreases this

will decrease the solubility rswnl^anything that increa-
vWvu^^ (jJÙ^^^ ^^^ orientation will increase the solMlity. \'
ky^c^iy^^^ The solute molecules^ therefore\'may be considered to be sur-
rhJp \'nbsp;rounded by more or less orientated water molecules. These will

not necessarily aU be similarly orientated. Thus the molecule
O\'^^^dUJiMiJnbsp;positive and negative polar groups. In general

Similarly if a neutral salt is introduced into the solution there
will be an attraction for one of the ions of the salt. Thus if there
is an excess of the orientation of type I (Fig. 7) the substances
will tend to attract cations, while if of type II it will attract anions.
In general, the substance will tend to attract that ion which orien-

tates molecules in the opposite direction, and repel the ion, which
orientates water molecules in the same direction. We will therefore
speak of cationphilic and anionphilic substances, according to
whether the substance displays an attraction for the cation or the

anion. These forces though not large will be sufficient to alter the
mean distribution of the ions and a cationphilic substance will have
in its immediate neighbourhood more cations than anions.

Now the ions, as has already been explained, will also have orien-
tated water molecules around them. The result will be that the
cations will bring to the cationphilic substance water molecules
orientated in a direction favourable to the solubility of the sub-
stance and so tending to increase its solubility. This may be re-
presented as in Fig. 7. (Type I),

The anion, on the other hand, by its opposite orientation will
tend to decrease the solubility, but its influence will be much less
on account of it being repulsed instead of attracted by the cation-
philic solute molecule. The net result of this quot;orientation influencequot;
will therefore be an increase in the solubility. As well as this orien-
tation influence there will of course also be a lowering of solubility
due to the fact that the hydrated ions quot;bindquot; a certain proportion
of the water molecules and so render them less free to take place in
the solution of the substance, as described on P. 15 (quot;salting outquot;
influence).

This influence, for an anionphilic substance, will also probably
be greater for the anions than for the cations. For a highly hydrated
ion, such as the SO4 ion, the salting out influence will preponderate
over the increase of solubility due to the orientation effect and con-
sequently the ion will lower the solubility — but for a slightly hy-
drated ion such as the CNS, the orientation influence will prepon-
derate and so there will be an increase of solubility.

Hence we will have a quot;spreadingquot; of the anions as is the case
with quinone. The cation influence will, as explained, be less marked,
but will nevertheless show itself in the lyotropic order, due to the
salting out influence.

If on the other hand we have an anionphilic substance (Fig. 7,
Type II) an exactly corresponding argument holds — but in this
case we will have a quot;spreadingquot; of the anions instead of the ca-
tions.

If then we examine the results as set out in Fig. 6, we must con-
dude that quinone, the nitro anilines and p-phenylenediamine
are anionphilic substances with positive polar groups which orien-
tate water molecules with their negative ends towards them (type

p-mtrophenol the last to a less marked extent) must be considered
cationphihc with negative polar groupsnbsp;^^laerea

The negative character of the acids may at first seem rather
su^nsing, but we must remember that as we are deahng^vith weak

lanty of these substances is correct in the recent work of Frumkin ^
from whose work it may be concluded that p-phenylenediamine and
quinone are (for our purposes) positive (anionphilic) and that
hydroquinone, boric acid and succinic acid are negative (cation-
phdic substances. (See p. 42). That such direct experimental con-
futation of the polanty of these substances is at hand is verv
fortunate as especially for quinone it would be difficult to arrive
at any conclusion through a chemical a priori reasoning

It wiU be seen from the figure that in the case of the nitro anilines
we have been successful in finding substances which give a spreading
similar to that for quinone - in fact the results for p-nitro aniline

amhne m eve^ way except that the NO, group has been replaced
by a second NH, group; and so we might expect a still greater
amon spreading. This expectation was quite justified by the results
the anion spreading being the greatest found. Further it is extre\'
mely interesting to find that we have actually a reversal of the cation
senes. Apparently the quot;salting outquot; influence of the cation has been
here so much reduced that lithium chloride actually increases the
solubihty. With the limited data available any detailed explana-
lon of this cannot be put forward; the explanation must however
rest in the factors already mentioned whose influence we do not
know quantitatively.

Here, for instance, we do not know how much of orientation in-
fluence and of salting out influence is contributed by the lithium

ion and the chlonne ion respectively; probably hoWever the whole
of the increase is due to the chlorine ion.

It may here be mentioned that if Linderstrom-Lang\'s theory that
the different behaviour of quinone and hydroquinone rested in the
fact that one was an oxidising agent and the other a reducing
agent (p. 19) was correct, we would expect that p-phenylenediamine
would show less anion spreading than p-nitro-aniline and that p-
nitro-phenol would show more anion spreading. Exactly the reverse
was however found.

p-Nitrophenol, as we expected, having no NHj group was found
to be less anionphilic.

The most markedly cationphilic substance was hydroquinone
(a benzene ring with two OH groups!) The anions are here extremely
close together. None of the salts used, however increased the so-
lubility. Probably Csl, would have done so.

Phenylthiourea seems to fall between the two types of substances.
This being the one solid substance for which a satisfactory cation
and anion series were formerly studied is then one reason why the
existence of cationphilic and anionphilic substances has not pre-
viously been noticed. It is difficult to understand why LiNOg is
here above KNO3. The possibility of other factors taking part in the
solubility influence must of course always be considered.

In spite however of these two or three irregularities in the series,
the results for these nine substances seem quite in agreement with
our main propositions:

That substances may be either quot;cationphilicquot; or anionphilicquot;, ac-
cording to how they orientate water molecules.

That the solubility influence of neutral salts consists of two parts (i)
ihc quot;salting outquot; influence — which is greatest for the most hydrated
ions — and (2) an quot;orientation influencequot; by which anions tend to in-
crease the solubility of anionphilic substances and cations the solubility
of cationphilic substances.

Frumkin ^ in a recent series of papers has described experiments
in which he has measured the potential difference at a water-air

mterface when the interface is charged by the presence of various
inorganic and organic compounds. In this way he was able to show
that methyl alcohol or acetic acid, for instance, charged the sur-
face positively, the more negative ends of the molecules (OH and
COOH respectively) being below the surface and the CH3 group
above the surface.nbsp;^ ^

On the other hand, if instead of acetic acid, trichloracetic acid is
taken, the onentation remains the same, but now the negative CI
atoms give a negative charge to the surface. Succinic acid in this
way gave a positive charge. In this case the negative carboxvl
groups are turned towards the water, as with acetic acid

Aromatic compounds, with a polar group charged the surface
positively though very much less than the aliphatic substances
Thus mtrobenzene gave a positive charge. o-Nitrophenol gave a
positive charge but m-nitrophenol gave a negative charge and
p^i rophenol still more negative. As the OH group of the phenols
would certainly be turned towards the water, this shows that the
1NO2 group IS also negative and so p-nitrophenol is (for our purpo-
ses) a negative substance.

Hydroquinone gave a negative charge, but quinone a positive.
These being both para compounds, it seems that the signs of their
charges must be the same as those of their respective polar groups
which fits m with our explanation of hydroquinone being cation-
philic and quinone anionphilic.

p-Phenylenediamine on the other hand gave a positive charge
to the surface. The NH^ groups are therefore positive, and so p-
phenylenediamine was rightly considered anionphilic

The nitro-anilines though not actually dealt with by Frumkin
would of course come between p-phenylenediamine and nitrophe-
noi, which is also in accordance with our results

For more details of these experiments the original papers should
be referred to. The evidence however seems quite convincing that
the groups OH, COOH, NO^ are negative while NH^ is positive
and that quinone and hydroquinone have positive and negative
polar groups respectively.

The Influence of Salts on the Sui^face Tension of Water
If a neutral inorganic salt is dissolved in water, the surface ten-
sion is increased. The surface tension of the solution may be ex-
pressed by the equation:

as = ow (1 mc)
where as is the surface tension of the solution, aw the surface ten-
sion of pure water, c the concentration of the salt and m a constant.
Actually, as was shown by Heydweiller ^ m is not strictly constant
but passes through a minimum at about 1 mol. The action of the
salt in increasing the surface tension is unmistakably an additive
effect of the ions composing it, as was first pointed out by Valson
The action of the different ions may be seen from the table given
below which is taken from Freundlich\'s quot;Kapillarchemiequot;, the re-
sults having been calculated from Heydweiller\'s ^ data:

\' Heydwcillcr Ann. d. Physik (4) 33 154 (1910).
\' Valson. Ann. de Chim. et de Phys. (4) 20 361 (1870).

Hiss \' showed that the dynamic surface tension of water was con
sjdembly greater than its static surface tension. In his ZeZTnl
the hqu.d under investigation was sucked into a capilbX mean
of a powerful a.r current to a height greater than tL dL fo^apS
lanty and atomi^d into spray. On stopping the air cu^ent a? a
defmite time the hquid rapidly sank to the position determTned bv
he surface tension. The position of the meniscus at veryTort L
te^als of fme was now determined. With benzene and nitroben ene
the static surface tension was established in the shortest time mea
urable under the experimental conditions (less than O.TOl squot; )
With pure water, however, the results are different, and an arprt
aablejime elapses before the value quot;of the static surface tenZ is

Transition from Dynamic to Static Surface Tension of Pure Water

bjn^^\' 7 from Freundlich\'s Kapillarchemie, the results having
been recalculated from Hiss\'s data.nbsp;^

tension are found to have a higher concentration at the surface
than in the bulk of the solution.

The exceptional properties of water (i.e. ^ its extraordinarily high
boiling point compared to its molecular weight 2, a specific heat
nearly twice that of all other liquids expansion instead of con-
traction on solidification, and what is absolutely unique, a con-
traction on heating between 0° and 4°C.) led Röntgen ^ to put for-
ward a theory that water consisted of a binary mixture of quot;water
moleculesquot; and quot;ice moleculesquot;, the latter of greater complexity but
less density. This theory was developed by Sutherland 2 who
suggested that steam consisted of molecules of HjO, ice (HjOjg, and
water a mixture of (H20)2 and (H20)3, the (HjOjj being Röntgen\'s
quot;waterquot; molecules. Bousfield and Lowry ® consider water a ter-
nary mixture of (H20)3, (H20)2 and HjO molecules, a theory which
seems more in accordance with the facts. They also give reasons\'
for thinking that water of hydration is in the form of (HjO) 2 mole-
cules. Our conception of water of hydration being orientated H2O
molecules tending to form chains is in some ways in agreement with
this idea. We can, however, go further with our hypothesis and assu-
me that the more complex quot;moleculesquot; present in water are really
chains of H2O molecules, these having a smaller density than the
single molecules. On lowering the temperature, more of these
quot;chainsquot; would tend to form (equivalent to more ice molecules).
The single water molecules will be found in larger concentration
in the surface of the liquid, and these having a lower surface tension
(equivalent to steam molecules) would account for the static sur-
face tension of pure water being lower than the dynamic surface
tension as shown in the experiments of Hiss. •

On the above hypothesis we may explain the influence of salts
on the surface tension. If hydrated ions are introduced into the
Water a certain number of these will be present in or immediately
below the surface, and will consequently introduce quot;water of hy-
drationquot; into the surface. This having a higher surface tension than
the simple molecules will increase the surface tension of the liquid,

\' Röntgen, Wied. Ann. 45 91 (1891).
\' Sutherland, Phil. Mag. (5) 50 460 (1900).
• Bousfield and I^wry, Trans. Faraday Soc. 7 85 (1910).
\' See note 3 page 41.

It is therefore true to say that the influence of salts on the surface
tens.on_of water depends on the salts effecting the equiHb^m

than once); this m its turn however ultimately depends on the
onentation of the water molecules by the salts

Superimposed on this purely quot;lyotropicquot; effect there may pos-
sibly be a second effect which is not due to the interaction of Xe
and solvent (i.e. a non-lyotropic effect).

The fact that water shows a maximum in the temperature-
dens.ty curve at 4X. is intimately connected with the to that

cifle of vanous salts per litre. Thus potassium chloride produces a
owering m the maximum density of 11.6X. The folloLHable
shows some of his results:nbsp;^

Molecular Lowering of the Temperature of Maximum Density of

and wp. .hu V- r , ^ quot;umoer oi determinations
a^as able to confirm the law of Despretz 3 that the lowering

produced by the addition of a solute is directly proportional to the
concentration of the solute. He also shows that the lowering pro-
duced by a highly ionised binary electrolyte is an additive effect
of the cation and anion, and that the molecular lowering can be
calculated by the addition of two moduli to the lowering produced
by a molecular solution of a chosen standard substance (e.g. hy-
drochloric acid).

The expenments were made by means of a dilatometer the
readmgs bemg accurate to about 0-2°C. It will be seen that the
molecular lowering is exactly proportional to the concentration
for each particular substance, a result which is rather surprising

When we consider in what order the ions exert their effect we
find for the cations:

NO3 gt; I gt; ^SO, gt; Br gt; CI
This IS not quite the lyotropic series. Na and SO^ are quite out
of their place, and NO3 does not come between I and Br On the
whob however thejess hydrated the ion, the greater seemsl^e
Its influence in lowenng the maximum density of water. Probably
we have here a lytropic effect superimposed on one or more other
effects. Thenbsp;lyotropic effect (due to a change in the equili-

bnum (H 0)„ _ nH,0 as already referred to under surface ten-
sion) would show itself in the density of the water of the solution
(as distinct from the density of the solution) rather than in a
shift of the temperature of maximum density. It is unfortunate
that there are not sufficient data available for constructing the
density-temperature curve for each salt solution. Such curves would
probably be very interesting.

Bousfield and Lowry ^ showed that on the addition of 2 % NaOH
the density maximum disappears; for 12 % solutions and upward
the curve has a simple parabolic form, while 42-5 % NaOH makes
the density temperature strictly linear from 0° to lOOX A compa-
nson of the minimum amount of such salts necessary to bring about
such effects would no doubt also show the lyotrppic series.

In table 20 are given the relative viscosities of various salt solutions
ot 1 normal concentration. In each case the viscosity of water at
the temperature of the experiment has been taken as unity.

The increase in viscosity is almost directly proportional to the
concentration of the salt for those salts which increase the vis-
cosity.

To these figures may be added that sodium salts increase the
viscosity in the order NaOH gt; Na2S04 gt; NaCl gt; NaBr gt; NaNOg

and acids increase the viscosity in the order NaCl gt; HjSOj gt;
HCl gt; HBr gt; HNO3. The alkaline earth chlorides increase the
viscosity more than the alkali chlorides in the order ta gt; Sr gt; Ba.

We see then that some salts increase the viscosity while others
lower it, the greatest increase being caused by the most hydrated
ion, the ionic series being

Li gt; Na gt; H gt; K gt; Rb gt; Cs
SO4 gt; tl gt; Br gt; NO3 gt; I
order of decreasing viscosity. We have then here again the
lyotropic series. This is what we would expect if as a very rough
^irst approximation we assume Einstein\'s ^ formula for the vis-
cosity of a liquid in which spherical particles are suspended:

rjs — -rim (1 2-5 9)
^l^ere vjg is the viscosity of the suspension, rjm that of the pure
^inid, and 9 the volume of the particles in unit volume of the
suspension. In other words if we assume that the greater the
Volume of the suspended particles (here hydrated solute) the

greater will be the viscosity, we must expect ^e_salts_composed
of the most hydrated ions to give the greatest viscosity The re-
sults show however that as well as this there musrbe~a^cOTd effect
tending to decrease the viscosity only overcome by the first effects
in the case of appreciably hydrated ions. A suggestion as to what
^^ this other effect is cannot be made without taking up the whole
theory of viscosity, which is outside the scope of this work.

The velocity of chemical reactions is also influenced by the pre-
sence of salts in the solution, and here again we find the lyotropic
series. With this phenomenon we find moreover that we get a
reversal of the lyotropic series in some cases according to whether
the action takes place in acid or alkaline medium. This was first
pointed out by Hober ^ Thus in acid catalysis of esters, and acid
inversion of cane sugar, bromides favour the reaction more strongly
than chlorides while sulphates inhibit it, the series being

Li gt; Na gt; K gt; Rb gt; Cs
On the other hand when we turn to the reactions in alkaline so-
lutions we find that sulphates = assist basic saponification of es-
ters while chlorides, bromides, nitrates and iodides inhibit to an
increasing extent, the series thus being

SO^ gt; Cl gt; Br gt; NO3 gt; I
for the anions, while Hober found for ester hydrolysis the cation
series to be

The whole subject of reaction velocity cannot be discussed here.
A possible explanation of tlie above facts however occurcd to us,
and since it follows more or less logically from the assumptions al-

ready made to explain the solubility influence it is perhaps worth
while putting forward, especially as up to the present no other
explanation has been suggested.

The velocity of the above mentioned reactions is governed by
the catalytic action of either hydrogen or hydroxyl ions as the
case may be. Lapworth ^ has put for^vard the theory that the
hydrated hydrogen ion has much less catalytic activity than the
unhydrated hydrogen ions. He examined widely differing examples
of hydrogen ion catalysis in organic solvents and showed that small
quantities of water produced a marked retardation in all cases.
Those reactions which occurred in both water and alcohol were
found to go enormously faster in alcohol. Lapworth accounts for
this by supposing that the water causes a reduction in the number
of nonhydrated hydrogen ions, which he supposes to be the active
catalytic agent. Dawson = from experiments on the reaction be-
tween acetone and iodine also supports this view. Lapworth assu-
mes the hydrated hydrogen ions are complex ions probably of the
form {HaO.H)^, the theory however may be easily adapted to fit
in with our somewhat different views on hydration. The chief point
is that anything tending to increase the hydration of the hydrogen
ions will decrease the velocity of the reaction while anything decreas-
ing the hydration will increase the velocity of the reaction. A similar
line of reasoning to that used in explaining the quot;orientation effect
which gives rise to an increase in the solubility will lead us to con-
clude that the anions will tend to increase the hydration of the hy-
drogen ions while cations will dehydrate. Consequently we md
that the most hydrated anion, SO,, has the greatest inhibiting
action, while the most hydrated cation (lithium) tends most to
increase the velocity of reaction. This will be the case for all re-
actions catalysed by hydrogen ions, but for a reaction dependent
on the catalytic influence of hydroxyl ions the twonbsp;^

reversed. sinL the cations will hydrate while the anions wi del y-
clrate, which is in accordance with the expenmenta data. This

be a tendency for the salts to increase the quot;active massquot; of the
reacting substances, (corresponding to quot;salting outquot; in solubility)
so accelerating the reactions, but this will not be so important as
the first influence and will not prevent the complete reversal of
the series.

The suggestion that the neutral salt effect in accelerating the
velocity of hydrogen ion catalysed reactions is due to the dehydra-
tion of hydrogen ions is put forward by H. S. Taylor \\ who points
out that this theory is in accordance with the fact that the neutral
salt action is independent of the substrat 2. The possibility of cer-
tain ions quot;hydratingquot; arises as a consequence of the assumptions
made in accounting for the increase of solubility in solubility in-
fluences.

\' H. S. Taylor, „A Treatise of Physical Chemistryquot;, P. 917.
\' Taylor, Medd. K. Vetensk. Nobelinst. 2 No. 34 (1913).

The cases we have described do not exhaust the phenomena
taking place in true solution in which the lyotropic influence of
salts is found. The adsorption of salts under certain conditions
may be cited as yet another example. Thus Od6n ^ found that the
adsorption of nitrates on charcoal took place in the following
order,nbsp;^

cause more work is necessary to withdraw the more hydrated
ions from the bulk of the solution, owing to their greater affinity
for the water and also because the protecting quot;sheathquot; of hy-
dration makes the adsorption more difficult. The adsorption is
of course often complicated by other influences. The experiments
of Kruyt and van der Madequot; and Kruyt and van der Spek ® show
that lyotropic influences play a part in the neutral salt influence
in dyeing. Such phenomena are however often too complicated to
be easily interpreted and will not be dealt with here.

We see therefore that there are a number of phenomena on
which the influence of neutral salts is due to the independent ef-
fects of the cation and anion which exert their influence in the
following order.

Li gt; Na gt; K gt; Rb gt; Cs
\'\'indnbsp;SO, gt; CI gt; Br gt; NO3 gt; I gt; CNS

\' Od6n and Anderson, J. Phys. Chem. 35 311 (1911); Oden and Langelius.
J- Phys. Chem. 25 385 (1921).

\' H. R. Kruyt and Miss J. E. M. van der Made, Proc. Kon. Akad. v. Weten-
sch. Amst. 20 636 (1917).

These are also the series obtained if we arrange the ions in
order of decreasing hydration. In the case of the solubility influence
we have shown that the lyotropic influence of the ions is due to two
reasons, firstly the hydration of the ions (giving rise to quot;salting outquot;)
and secondly the orientation of water molecules (which sometimes
gives rise to increased solubility). The second reason accounts for
the occurence of a specific influence of either cation or anion. Some-
times these two effects act in opposition to one another, which may
give rise to an irregularity or reversal of one of the series. These
properties of the ions also account for their influence on other
lyotropic phenomena. In some cases, (e.g. influence on surface
tension of water) we have seen that the salts influence the equili-
brium (H20)n:;!:nH20but this influence is really dependent on the
quot;orientationquot; influence of the ions.

Note. The conception that ions orientate water molecules is
of course not new. Thus Fajans ^ postulates a similar quot;polarisationquot;
of the water molecules by the ions to account for the solubility of
alkaline halides being least when the cation and anion are about
equally hydrated.

If we determine the minimum concentration of a number of
salts necessary to bring about the flocculation of a typically lyo-
phobic sol we find, as is well known, that this minimum con-
centration (hereafter referred to as the flocculation value or F.V.)
varies very considerably with the valency of the ion of opposite
sign to that of the charge on the sol particles. Thus for the AsjSs
sol we find the following flocculation values as determined by
Freundlich ^

AICI3 0-093 „ „ „
The dependence of the flocculation value on the valency of the
ion is fully dealt with in text books on colloid chemistry » and will
not be considered here. If however, instead of taking cations of va-
rious valency we take cations of the same valency, we find again
differences for the F.V.s. though certainly very much less marked
LiCl 58 millimols per litre
NaCl 51 ,, M
KCl 49.5 „ „
Here again we have the lyotropic series. The stability of lyopho-
bic sols is governed by the charge of the particles. The stability of
lyophilic sols depends partly on the charge of the particles but also

\' Freundlich Z. phys. Chem. 44 129 (1903). Ibid 73 385 (1910).
\' Freundlich. quot;Kapillarchemiequot;; H. R. Kruyt. quot;Colloidsquot;: etc.

We might therefore expect that the lyotropic influence of salts
would be more marked in the case of lyophihc sols, and this in
fact IS the case. Thus with the agar sol (which may be considered
the typical lyophilic sol as de Jong ^ has pointed out) we find that

can be prepared. This quot;salting outquot; is undoubtedly closely related
to the quot;saltmg outquot; influence of salts on the solubility of substances
and mvolves a dehydration of the hydrated particles De long
has shown this to be the case by means of viscosity measurements
mth the agar sol, additions of MgSO, in concentrations just less than
that required to salt out bringing about a marked drop in the vis-
cosity. In general then, we find that the more hydrated the ions
of the salt, the more easily will that salt quot;salt outquot; a lyophilic sol
In this phenomenon it should be remembered, quite large concen-
trations of salt are necessary to bring about flocculation, the phe-
nomenon being in fact quite different to that of the flocculation of
lyophobic sols where very small quantities of salt can bring about
complete flocculation - in the case of the lyophilic sol not only has
the charge of the particles to be removed but also their hydration
There are however, a number of sols which in a sense stand be-
tween the lyophilic and lyophobic. Though not truly lyophilic
sols they are without doubt to some extent hydrated. Amongst
these may be mentioned the ferric hydroxide sol and the alu-
mimum hydroxide sol (both positively charged) and the vanadium
pentoxide sol (negatively charged). Flocculation values for all these
sols (as will be shown later) show a much more marked lyotropic
mfluence than is to be found from the flocculation values of ty-
pically lyophilic sols.

Scarpa ^ has shown that a lyophilic sol (such as agar or gela-
tin) can be dehydrated by adding a sufficiently large concentration
of alcohol, the resulting alcoholic sol having all the properties of a

lyophobic sol — low viscosity, sensitivity to electrolytes, particles
visible in ultramicroscope etc. If then the appearance of a strongly
marked lyotropic influence of salts on a sol, as shown by the rela-
tive values of the F. V. for different salts, is connected with the hy-
dration of the colloidal particles, dehydrating such sols with alco-
hol should remove the influence. We therefore thought that the
determination of the F. V.s for various salts in various concen-
trations of alcohol should throw some light on the mechanism of
this lyotropic influence. The results of such experiments are given
in the following sections.

Freundlich and Leonhardt ^ have made a study of the VjOg
and the MojOg sol and the influence of electrolytes upon them. In-
stead of determining flocculation values they determined turbidity
values, that is the amount of salt required to bring about a marked
turbidity within a definite time. The following are their results for
the VjOj sol prepared according to the method of Biltz.

A similar remarkable spreading of the lyotropic series is shown
in their results for the MogOj sol.

The fact that both of these sols give gelatinous precipitates and
that the VjOj sol readily sets to a gel may be taken as suggesting
that the sols are to some extent hydrated and this is borne out
^y the results of our experiments.

The sol was prepared from ammonium vanadate according to the
method of Biltz ^ sUghtly modified. The ammonium vanadate was
rubbed up with dilute hydrochloric acid in a mortar, a red powder
separating out from the mixture. This was washed several times by
repeatedly mixing with water, allowing to settle in a glass cyhnder
and decanting. After this had been repeated seven times a point
was reached where there was no longer settling. The whole was
then transferred to a large flask, made up to four litres with distilled
water and shaken up vigorously until thoroughly peptised. A very
small amount of thymol was added to protect the sol against bac-
terial contamination. It was also found that the sol lasted longer if
kept in the dark.

This sol was not precipitated on the addition of an equal volume
of 96 % alcohol even after standing for two or three days. The co-
lour of the sol with alcohol added was however, much browner than
the original sol which was more of a reddish colour. This colour did
not appear immediately after adding the alcohol but developed
slowly over a period of about five hours. The browner alcoholic sol
exhibited a brighter cone when viewed through the slit ultrami-
croscope. A higher concentration of alcohol (15 ccs. of alcohol ad-
ded to lOccs. of sol) brought about complete precipitation after
the three and a half hours.

The flocculations were carried out in flat bottomed glass tubes
(2 cms. in diameter and 10 cms. high). The tubes were tightly closed
by corks covered with pure tin foil.

Flocculation values for lithium, sodium and potassium chlorides
were first determined without the addition of alcohol. For this pur-
pose the sol was diluted with an equal volume of water. lOccs. of
this diluted sol was pipetted into one of the glasses and 5ccs. of
the salt solution of thi-ee times the required final concentration was
added by means of another pipette and the contents of the glass

thoroughly mixed by shaking. The figures given for the salt con-
centration in the tables in all cases refer to the concentration in the
final mixture of sol plus salt solution. Several glasses were thus filled
with increasing concentrations of salt and allowed to stand for
two hours, at the end of which time the concentration just suffi-
cient to bring about flocculation was noted. The sol was considered
to be flocculated when a clear zone was visible between the me-
niscus and the dark red brown precipitate. The flocculation value
having been thus roughly determined another series was made
giving a more accurate value and so on until the correct value was
reached as accurately as the experimental error would allow. The
F.V.s could be determined correct to about 4 %; with the alcohol
mixtures the error was slightly more, due to the settling of the floc-
culation in the presence of alcohol not taking place so uniformly (in
some cases a precipitate would float at the top instead of sinking).
The results were however in all cases sufficiently accurate to leave
no doubt about the phenomena here described. The flocculation
values in the presence of the alcohol were carried out so that the
concentration of the VjOb was the same in the final mixtures as
in the flocculation values without alcohol. Thus for the addition of
% by volume of alcohol, 33J ccs. of alcohol and 1ccs. of
water were added to 50ccs. of the original sol. lOccs of this mix-
ture was then placed in the glass. A salt solution was then made up
containing the same concentration of alcohol, and three times the
concentration of salt required in the final mixture. 5ccs. of this al-
coholic salt solution were then pipetted into the glass as before.

By this means, all the flocculations were carricd out on a sol of
the same concentration of vanadium pentoxide and the addition
of the salt solution was not accompanied by the disturbing effects
which would arise on mixing if a salt solution was added of different
alcohol content to the sol.

The results for LiCl, NaCl and KCl for the aqueous V^Os sol and
the sol plus two different concentrations of alcohol are shown in
the table. The experiments were carried out as quickly as possible
so that the error due to the change of flocculation values with time
^vould not be appreciable. The final determinations for all salts at
each particular concentration of alcohol were carried out simulta-
neously.

LiCl .
NaCl.
KCl .

The flocculation values are given in millimols per litre. It will be
seen then that for the aqueous solution we obtained results similar
to those obtained by FreundUch and Leonhardt. The addition of
alcohol however as well as making the solution very much more
sensitive to all three salts (as might be expected if the stability of
the solution is to some extent due to its hydration) brings the floc-
culation values relatively as well as absolutely, much closer together
Thus while for the ordinary sol the F.V. for LiCl is over twenty
times that for KCl, the addition of an equal volume of alcohol makes
the F.V. for LiCl only i:7 times that of KCl.

The precipitate formed in the alcohoUc sols differed in appearance
for the different salts. Thus the precipitate with KCl was much more
voluminous than the precipitate given with NaCI, and that with
NaCl more voluminous than that given with LiCl. The colours also
differed, the LiCl precipitate being brownish like the colour of the
alcoholic sol while the KCl precipitate was more reddish. On shaking
up the precipitates and allowing them to settle again the LiCl pre-
cipitates settled rapidly in about three minutes The NaCl precipi-
tates took about twenty five minutes and the KCl ones over an
hour.

The ferric hydroxide sol was chosen as an example of a positively
charged sol exhibiting somewhat lyophilic properties.

dissolved in 1 litre of water. A solution of ammonium carbonate was
added slowly to this while stirring continuously until the point was
reached where there was just no remaining precipitate. The solution
was then dialysed on a Zsigmondy-Heyer star dialyser for three
days, a freshly prepared collodion membrane being used. At the
end of this time the water showed no reaction for chlorine ions with
silver nitrate. At the end of the dialysis the volume of the solution
had increased to about one and a half litres.

These were carried out similarly to the determinations with the
vanadium pentoxide solution.

2ccs. of the sol diluted with water or alcohol were placed in the
flocculation glass and then Icc. of the salt solution of the same al-
cohol content as the sol was added to this and the whole well mixed.
The tubes were tightly closed with corks covered with tin foil and
allowed to stand for 24 hours, after which the salt concentration
which was just enough to bring about flocculation was noted. The
sol was considered to be flocculated when there was a clear zone
of liquid between the meniscus and the precipitate. The longer
period (24 hours) before reading was found more suitable for these
experiments since the precipitate formed in alcoholic sols settled
niuch more slowly than in the aqueous solution, and consequently
if the reading were made after a shorter time (one or two hours) it
Was impossible to compare the limit values for the two sols.

Experiments were made with KCNS, KI, KNO3 and KCl and
the results are shown in the table.

It will be seen that the alcoholic sol is much more sensitive to all
the salts in the presence of alcohol. On the other hand, the relative
values for KI, KNO3 and KCl are not much changed.

Unlike the vanadium pentoxide sol, the ferric hydroxide sol was
quot;ot precipitated by alcohol alone, even when added in very large
quantities, e.g. ten volumes.
The two F.V.s for KCNS for the aqueous solution were obtained
the beginning and end of the experiments respectively and show
tl^at the solution had not changed appreciably during this time.

Experiments with Aluminium Hydroxide Sol
This was chosen as another example of a positively charged sol.

The sol prepared by the method of Gann^ is said to gelate on addi-
tion of electrolytes instead of flocculating and consequently a sol
was prepared by another method described by Müller

A preparation of crystalline aluminium chloride of the quot;Onder-
linge Groothandelquot; was used. This was found to have only a trace of
iron present as shown when treated with potassium thiocyanate.
100 gs. of the aluminium chloride were dissolved in about 750 ccs.
of water. 75ccs. of this was diluted to about 250 ccs. and precipitated
by adding dilute ammonium hydroxide to the boihng solution until
there was a slight extess. The solution was then filtered and the
precipitate washed three or four times as quickly as possible with
hot distilled water. The precipitate was then transferred to a beaker
with about 600 ccs. of water and boiled. O.IN HCL was added. 2 ccs.
at a time, the contents of the flask being boiled for a few minutes
after each addition, this being continued until the amount of hy-
drochloric acid necessary to peptise the sol had been added. The last
two or three cubic centimetres of the acid was added in still smaller
portions.

In this way an opalescent, rather opaque sol was obtained. Ana-
lysis showed it to contain 0-424 grams of Al^Og per litre. A slight

\' Gann, Kolloid Chem. Beihefte 8 64 (1916).
\' Zeit, anorg. Chem. 57 310 (1908).

precipitate was formed from the sol during the next forty eight
hours. The sol was consequently decanted before being used for
the following experiments.

The sol was found not to be flocculated even on the addition of
very large quantities of ethyl alcohol (seven volumes of alcohol
to one of sol). On the other hand acetone flocculated the sol on the
addition of three volumes.

The flocculation values were carried out exactly as in the case
of the ferric hydroxide sol. The readings were made after twenty
four hours, the sol being considered flocculated if any precipitate
had formed at the bottom of the glass. This was found to be the
most suitable end point as in some cases the flocculation was not
complete.

Flocculation values were determined for KCNS, KI, KBr and
KCl. The corresponding determinations for KNO3 could not be made
owing to the small solubility of this salt in alcohol being too low.

When a figure such as quot;gt; 850quot; is given this means that this
was the highest concentration of salt tried and that at this con-
centration there was no flocculation, it being impossible to reach the
flocculation value on account of the limited solubility of the salt
in the alcohol concentration used. The figures given in the last co-

lumn are not strictly comparable as the concentration of the sol is
here somewhat different.

The check determinations made with KCl at the end of the ex-
penments (after eight days) show that there was no appreciabk
expenmental error due to the changein thestabiHty of this with time
Before offenng an explanation of the phenomena described for
these three sols of the somewhat lyophilic type we will first describe
some similar experiments carried out with both a negative and a
positive typically lyophiHc sol - namely the gelatin sol. In these
expenments it will be seen we obtained direct evidence that an ion
of opposite charge to the charge of the sol can under certain circum-
stances actually stabilise the sol by means of its water of hydration
a fact of pnmary importance in explaining these results.

For this method isoelectric ash free gelatin (P„ = 4-7) was
prepared by Loeb\'s method from the best quality photog apWc
gelatin produced by the quot;Lijm en Gelatinefa^kquot; o? Delf^

A I /o gelatin sol was made up. For this purpose the weighed
quantity of gelatin was allowed to swell for one hour in cold wfter
It was then heated on a water bath to 60°-65\'\' until dissolved and
subsequently filtered hot through a thoroughly washed asli free
inter, the temperature not being allowed to fall below 40° So as
to preserve the solution a very small crystal of thymol was added
]ust after the gelatin was completely dissolved but before filtering

On adding a large quantity of alcohol to the isoelectric sol the
gelatin was immediately precipitated as an amorphous white mass
io fmd öxactly the least quantity of alcohol which would bring

about precipitation it was necessary to set out a series of mixtures
containing various quantities of alcohol but the same concentration
of gelatin. Thus a 75 % (by volume) alcohol mixture was prepared
by adding 30ccs. of 96 % alcohol to lOccs. of the 1 % gelatin sol
(the mixture thus containing 0-25 % gelatin) and a 30 % alcohol
sol was prepared by adding 18ccs. of water and 12ccs. of alcohol to
lOccs. of gelatin sol. Intermediate concentrations were obtained si-
milarly. 5 ccs.of these various mixtures were placed in flocculation
glasses and examined after having stood all night:

precipi- precipi- precipi- tated
tation tation tation immediately
Thus 42 % alcohol was the smallest concentration to bring
about a precipitation. 75 % alcohol brought about immediate
precipitation. Concentrations between these values brought about
opalescence which was followed by precipitation on standing.
40 % alcohol and below gave no opalescense and no precipitate.

The addition of alcohol even in very large quantities (e.g. ten
volumes of alcohol (96 %) to one of sol) did not cause a precipi-
tate. Mixtures containing increasing quantities of alcohol were set
out in flocculation tubes exactly as with the isoelectric sol, the ge-
latin concentration (0-25 %) being the same in all of them.

The following concentrations were tried
50 52-5 53-75 55-0 57-5 60-0 75-0 ^•^^fJJTaSr\'
53-75 % alcohol or over caused the sol to become opalescent,
50 % alcohol or even somewhat less caused a slight opalescence
^fter standing for some hours. In no case was there a precipitate
formed. It was therefore expected that the sol with about 50 %
alcohol would be sufficiently sensitive to electrolytes for the pur-
pose of our experiments.

The flocculation values for the various salts were determined in a
similar way to the methods already described for the other sols.
In every case the sol was diluted four times, by adding to one vo-

lume of sol 3 volumes of a mixture of alcohol and water so that the
final mixture contained 0-25 % gelatin and the required percentage
of alcohol. 2ccs. of this mixture were pipetted into a number of
flocculation glasses and to each glass was added 2ccs. of a mixture
of electrolyte solution and alcohol containing the same percentage
of alcohol and twice the concentration of electrolyte required in
the final mixture of 4ccs. The glasses were well shaken immediately
after the addition of the electrolyte and allowed to stand for about
18 hours, being well corked with corks covered with pure tin foil.
After this time it was noted which concentrations of salt had caused
a precipitation and which had not.

If the influence of alcohol on gelatin was similar to that on the
vanadium pentoxide sol it would be expected that high concen-
trations of alcohol would give flocculation values for KCNS, KBr and
KCl which did not differ from one another by much. Some preliminary
experiments were therefore made with 75 % alcohol which gave the
following results.

The Existence of an upper limit to the salt concentration necessary
to flocculate

With 60 % alcohol however some results which had not been
expected were obtained. Here it was found that not only was there
a limiting value for the electrolyte below which no precipitate was
formed, but that there was also a second limiting value above which
no precipitate was obtained. This will be seen in the widely spaced
series for KCNS (60 % alcohol) shown in table 24.

It will be noticed that for the higher concentrations where no
precipitation was obtained, the opalescence on the other hand ac-
tually disappeared. This at once made us think that there might
be some connection between the existence of this hitherto unex-
pected upper limit value and some observations made by Scar-
pa Scarpa found that for a gelatin-water-alcohol system con-
taining 48-7 % alcohol by weight, some electrolytes caused pre-
cipation while others brought about no precipitation and actually
caused the opalescence of the sol to disappear. He also found that
the addition of solid sodium chloride did away with the opalescence.
De Jong found that, if after the addition of the solid sodium chloride
some more alcohol was added, the opalescence returned. He there-
fore put forward the suggestion that this was a lyotropic pheno-
menon quot;in which the viscosity maximum of the binary system alco-
hol-water was displaced.

To return to our own experiments, we found that the addition of
niore alcohol to the mixture containing 1360 mm. of KCNS caused
the opalescence to return. We also found, on adding solid sodium
chloride to the original 60 % alcohol gelatin sol, that, as found by
Scarpa, the opalescence disappeared and no precipitation was
brought about even by dissolving quite a considerable quantity of
the sodium chloride. Further, solid potassium thiocyanate was found
to have the same effect.

Some very interesting results were obtained by placing the dry
crystals of the salt at the bottom of a flocculation tube, carefully
pouring the alcoholic gelatin sol over the crystals and allowing the
salt to diffuse upwards through the sol. As the salt diffused up-
wards through the opalescent sol, there first appeared a zone of
^uite clear sol immediately above the crystals, sharply defined
from the still opalescent sol above it. As diffusion proceeded this
2one extended upwards. Later a second clear zone (corresponding
to the higher concentrating of electrolyte above the higher limit
^alue already mentioned) appeared and extended downwards. This
^PPer zone was separated from the opalescent zone by a jagged
^^ge formed by the precipitate from the upper part which floated

\' Scarpa — Koll. Zeit. XJ 8 (1914). cf. also de Jong Diss., Utrecht (1921)
p. 85,

on the opalescent sol. The bottom clear zone on the other hand
continued to be sharply defined by a horizontal line of demarca-
tion. If not too great a quantity of salt was used, it was eventually
all dissolved, and it could then be clearly seen that there was no
precipitation whatever in the bottom zone. Here then we have
the various stages of the irregular series (p. 66) akeady described
present in one test tube: at the bottom, (highest concentration of
electrolyte) no precipitate and opalescence destroyed, i.e. a zone
where the sol is actually stabilised and seems to be more lyophilic,
in the middle a zone where there is no precipitation but the opales-
cence remains (lower concentrations of electrolytes), and at the
top a zone where precipitation has taken place leaving a clear li-
quid (still lower concentrations). The concentrations which are too
low to bring about a precipitate are present at the top of the tube
eariier in the experiment before diffusion has gone far enough to
allow a zone of precipitation to form.

We found these phenomena could be repeated with more or less
ease for any combination of electrolyte and alcoholic concentra-
tion which give a lower and higher limit value (see later).
We may here give a description of two actual experiments.
A gelatin sol containing 55 % alcohol was used. This sol which
had been mixed 24 hours previously was strongly opalescent. Two
flocculation tubes were used, in (1) more sodium chloride crystals
were placed than could dissolve in the sol under the conditions des-
cribed, while the bottom of (2) was only just covered with crystals.
20ccs. of the sol was then run in slowly from a pipette.
This gave a column of liquid 75mms. high. A millim-
eter scale was placed behind each of the tubes and
they were allowed to stand untouched for a couple of
days, care being taken that they were well corked so
that there could be no evaporation of alcohol. The
results are given in Table 25.

In each case the top clear zone was separated from
the middle opaque zone by a jagged line, not horizontal,
which showed clearly the presence of an actual precip-
itate floating on the sol below. This is shown in Fig. 8.

Similar results were obtained with KCl and KCNS
dissolved too rapidly to give sharp results.

Let us consider a gelatin sol containing.sufficient alcohol to make
It opalescent and having the properties of a lyophobic sol as des-
cribed by de Jong, the gelatin particles being quot;dehydratedquot; by
the alcohol. If now some electrolyte is added, the ions of the elec-
trolyte of opposite charge to the particles will be attracted to the
surface of the particle, in a quantity depending on the concentration
pf electrolyte. If sufficient electrolyte is added, eventually a point
IS arrived at where the critical potential of the sol is reached and the
flocculates (the first limit value of our irregular series). The floc-
^nlation only takes place after the sol has been allowed to stand
for some time. If instead more electrolyte is added, a further increase
m the quantity of ions in the vicinity of the particles will take place.

These ions, hydrated in aqueous solution, will be surrounded by
a larger concentration of water molecules than exists in the
bulk of the alcohol-water system. Consequently by concentrating
round the gelatin particle they will increase the percentage of water
in its immediate neighbourhood, and so, if the quantity of alcohol
present is not too much they will bring about a quot;rehydrationquot; of
the gelatin. When this rehydration has taken place, the sol, al-
though no longer stable on account of its charge, will not flocculate
owing to the stabilising influence of the hydration of the particles.
This explains the second limit value above which no concentration
of electrolyte will bring about flocculation.

We now prepared a new gelatin sol and investigated both these
limit values at various concentrations of alcohol. The results are
summarised in the table. The first figure in the brackets is the high-
est concentration in millimols of the particular electrolyte which
would give a precipitate, the second figure being the lowest con-
centration to give a precipitate. The quot;end pointquot; for the determi-
nation of the limit values was in every case taken as the concentra-
tion which would bring about any precipitate, irrespective of size,
in some cases the flocculation being incomplete.nbsp;^

The experiments with the 40 % alcohol mixture were not so
accurate as the other, it not being possible to find the limit value
so sharply.

The significance of. these figures will be more appreciated Jjy
studying the graph (Fig. 9) in which percentage of alcohol is the
ordinate and concentration of electrolyte the abscissa. If we first

consider the points for the higher and lower limit values for KCl
and KCNS at 55 % alcohol, we will see that if the curves for the
limit values (both higher and lower) of KCl and KCNS respecti-
vely are to be continuous they must cut at some point. This is
found to be the case for the lower limit values between 45 % and
50 % alcohol. In the table this is shown by the reversal of the
series for the lower values between these two concentrations of
alcohol.

20 JO 40 SO OO 70 00
Conccnirulion of Electrolyte tn MilUmol»

We see then that in the figure we have an area of precipitation
corresponding to certain concentrations of alcohol and of salts.
Outside this area no precipitation can take place. To the right of
this area is the area where, although the concentration of alcohol
is high enough (about 39 %) to make precipitation possible, the
concentration of the salt is sufficiently high to bring about stab-
ilisation by reliydration at tlie particular concentration of alcohol.
\'I\'he most highly hydrated salt (KCl), it will be noticed, brings
about this rehydration most easily. In the lower flocculation values
it will be noticed that the sol is most sensitive to the most highly
hydrated anion, as are the aluminium hydroxide and the ferric hyd-
roxide sols.

We now prepared a new negative gelatin sol (as described on
page 77) and carried out similar experiments. The results are given
m the accompanying table.

With less than 70 % alcohol it was impossible to obtain precipi-
tates at any concentration.

With 55 % alcohol a sol was obtained that was distinctly opa-
lescent. This was aUowed to stand over solid KCl for two days, as
in the experiments with the positive gelatin. The bottom 7cms be-
came quite clear, but no flocculation appeared at top of tube, sug-
gesting that no flocculation would take place at any concentration
of KCl. This was further confirmed by the following series:

With 70 % and 75 % alcohol higher and lower limit values were
obtained as already shown in Table 27. For both these mixtures,

standing over NaCl gave the clear zone at bottom of tube as well
as flocculation at the top of the tube, leaving a cloudy zone in the
middle, as already described for the experiments with positive ge-
latin.

If we examine the results as they appear when set out in the
graph (fig, 10), we see that as in the case of the positive gelatin sol,
there is an area within which precipitation takes place and outside
which precipitation cannot take place. The curves for LiCl and KCl
do not cut as do the curves for KCl and KCNS in the case of the
positive sol. For the lower flocculation values the partially dehy-
drated sol is most sensitive to the least hydrated salt (KCl) as is the
case with the negatively charged vanadium pentoxide sol but not
with the positively charged sols. These and other facts will be dealt
with in the explanation which follows.

To explain these series for the precipitation of lyophilic sols by
salts, we will first deal with the negative sols, such as the partially
lyophilic vanadium pentoxide sol and the truly lyophilic negative
gelatin sol; this being a somewhat simpler case than that of the
positive sols. There are two outstanding facts: firstly, that there is a
very considerable lyotropic influence (especially in the case of the
vanadium pentoxide sol) as shown in the marked differences in the
flocculation values for salts with different cations and the same
anions, and secondly that the sol is most sensitive to the least hy-
drated cation (potassium). In explaining the solubility influence of
salts we showed that an ion might make its influence felt in two ways,
firstly by a stabilisation effect due to its power to orientate water
molecules and secondly, due to a quot;salting outquot; effect. If we consider
the mechanism by which a crystalloid is held in solution in a satu-
rated solution, we see that every molecule of the solvent plays its part
in keeping this quantity of crystalloid in solution; and consequently
if any of the water molecules are used for another purpose, as is
the case in salting out, some of the crystalloid must come out of
solution — in other words there is a decrease in tlie solubility. In
the case of a sol, however, every molecule of the water present
does not play a direct part in keeping the disperse phase in suspen-
sion — consequently a considerable quantity of a salt can be added
(especially if the sol is not greatly hydrated) without any quot;salting
outquot; taking place. In some cases (e.g. the agar sol) where a neutra-
lisation of the charge is not in itself sufficient to bring about floc-
culation, eventually a point will be reached (if the solubility of the
salt is high enough) where quot;salting outquot; of the sol will actually
take place., This case will be referred to again on page 79. But in
the case of a sol which, though to some extent hydrated, owes its
stability chiefly to its charge, no salting out effect will make itself
felt at the low salt concentrations which is sufficient to bring about

flocculation. The other effect, stabiUsation, due to the favourable
orientation of the water molecules by the ion of opposite charge,
will, however, appear. That is to say, the negatively charged vana-
dium pentoxide particle will tend to have its hydration increased
by the cations which neutralise its charge, the degree of this hy-
dration depending, of course, on the degree of hydration of the
cation. Consequently the hydration will be increased by the Hthium
ion, and the critical potential at which flocculation takes place will
be lower for LiCl than for KCl. Consequently, the highest floccu-
lation value is for lithium chloride, the series being:

Li gt; Na gt; K gt; Rb gt; Cs
instead of the other way round as would be the case if the salting
out effect played any appreciable part.

Evidence that this theory is correct is obtained from the experi-
ments witli the partially dehydrated gelatin. Here it was shown that
the addition of sufficient electrolyte, considerably in excess of the
minimum amount necessary to bring about flocculation, actually
re-hydrated the sol, this re-hydration being manifested by the
sol remaining stable with the higher concentrations of electroly-
tes and actually becoming water clear although the alcoholic sol
without the addition of the electrolyte was opalescent or opaque.

This, then, is the explanation of why we find that all negative
sols, such as the lyophobic sols of gold, platinum. As,S3, the semi-
lyophilic solutions of VA. Mo A. Sulphur (Syen Oden s) and the
truly lyoi)hilic sols such as agar and negative gelatin (the two atter)
When partially dehydrated by alcohol) are eas sensi ive to the
most highly hydrated cation, the series for the flocculation values
being:

Li gt; Na gt; K gt; Rb gt; Cs
(The aqueous sols of agar and neg^ive f^^^^^ quot;.^n
Hydrated is to be stable even when sufficientnbsp;yte ha b^^^^^^

•-iclded to entirely neutralise the charge; and consequently precip
tation [::::^tgt;ught about by salting follo^ng on^^n^^
nation of charge, which means that they will be most sensitive

Provided nocculation can be brought about by he
tl^o potential 5 to a critical potential, and that t - ^
in itself sufficient to leave the sol stable, this lyotropic influence

will be more evident the greater the hydration. Thus in the lyopho-
bic sol and the dehydrated Y^O^ sol it is quite small, but for the
aqueous VgOs sol the hydration of which seems to be considerable
the difference between the F.V.\'s for the various monovalent cations
is very large.

We will not here go into the question of what determines whether
a sol is by nature lyophilic or lyophobic, but we may assume that a
truly lyophobic sol is incapable of appreciable hydration even when
in contact with highly hydrated ions of opposite charge, so that the
lyotropic influence is always small. On the other hand, a sol such
as VjOg or the partiaUy dehydrated gelatin sol will, on account of
the peculiar structure of the particle and the peculiar surface forces
characteristic of this particle, tend to have its hydration increased
when in contact with highly hydrated ions of opposite charge and
this the more readily the greater the hydration of the particle under
ordinary conditions; so that lyotropic influence will here be evident.

It should be remembered that the decreased lyotropic influence
with the alcoholic vanadium pentoxide may be due not only to the
sol being made more lyophobic by the alcohol but also to the in-
fluence of the alcohol in dehydrating the ions of the added salts.
This dehydration of the salts by the alcohol may be the reason why
the lyotropic influence was never so marked with gelatin as with
vanadium pentoxide, as alcohol had to be added in a concentration
which was no doubt sufficiently high to appreciably dehydrate the
ions of the salts before the gelatin became sufficiently sensitive to
enable flocculation values to be obtained.

The case of the positive sol is somewhat different. Here wc see
that the sol instead of being most sensitive to the least hydrated
ion of opposite charge (CNS), is more sensitive to the more hydrated
ions, the flocculation values being in the order:
CNS gt; I gt; Br gt; NO3 gt; CI

On first sight it might be thought that this was evidence that
the quot;salting outquot; influence of the ions in some way came into
play much more than in the case of the flocculation of negative
sols by cations of the alkali metals. The lithium ion. for example
IS certainly more hydrated and not less hydrated than the chlorine
ion and further not only is there no evidence that positively charged
sols are more easily dehydrated than negatively charged sols but

on the contrary all the evidence goes to suggest that positively
charged sols are more difficult to dehydrate (for example the posi-
tive eerie hydroxide sol)

We must then look for the explanation in some other factor.
Is this factor to be found in the properties of the positive sols or in
the properties of anions as distinct from cations? No factor is to
be found in the properties of the sols themselves that does not have
its counterpart in the negative sols. If on the other hand we consider
the properties of the salts of the anion series we notice one distinct
difference from the cation series. Wliile for the cation series

LiCl gt; NaCl gt; KCl gt; RbCl gt; CsCl
solubility decreases with decreasing hydration of the cation, for
the anion series

KCl gt; KBr gt; KNO3 gt; KI gt; KCNS
solubility increases with decreasing hydration of the anion. We have
then in this anion series an increasing tendency to go into solution
running antibatic with the hydration of the anions. Thus, as we
go from the chlorine ion to the thiocyanate ion we find an increasing
quot;freedomquot; of the ion which can increase the solubility of the salt in
spite of the decreasing hydration. This same quot;freedomquot; of the ion
will, when the ion is adsorbed on the surface of the colloid particle,
increase the stability of the sol by imparting to the particle a kind
of buoyancy — in other words this is a peptising effect which is a
minimum with the chlorine ion and reaches a maxnnum with the
tliiocyanate ion. Such a tendency exists no doubt also with cations,
but would not make itself seen as it would run in the same direction
(synbatic) with the increasing hydration of the cations With the
anions, as we have shown, however, it runs antibatic to the hydra-
tion and is sufficiently great to bring about
ries with the flocculation of positive solutions. The
of this effect is no doubt also partly due to the
Crated than the cations with which they are compared in the expe

quot;quot;rlhis connection it should be remembered tl-t i^rked
peptising influence of iodine and thiocyanate 10ns is well knoNsn.

Thus von Weimam i has found that LiCNS and Lil are most active
in peptising cellulose (see p. 81).

This then may be taken as the explanation of why the flocculation
values for the aqueous positive sols (ferric hydroxide, aluminium
hydroxide, dehydrated gelatin) exhibit the lyotropic series in the
reverse order to the negative sols. The fact that the lyotropic in-
fluence does not become less marked in these sols on the addition of
alcohol (in the case of aluminium hydroxide it actually increases)
may be assumed to be caused by complications arising out of the
superimposing of this additional factor on those already mentioned.
As long as we are ignorant of the relative part played by these
factors an exact explanation must be impossible.

Lillie ^ describes the strong lowering of the osmotic pressure
of protein sols by electrolytes and finds an unmistakable connection
with the lyotropic series of the anions. This effect of the electro-
lyte may be due to its bringing about an aggregation of the pro-
tein particles into larger groups as well as its influence on the ap-
parent osmotic pressure through the effect on the membrane equi-
librium. There seems however to be no direct evidence to show that
this change in osmotic pressure is actually connected with a change
in particle size.

The appearance of the lyotropic series in the salting out of lyo-
philic sols was first shown by Hofmeister and his pupils ® who de-
termined the lower limit of concentration at which a sol of white
of egg was immediately rendered turbid. The greatest effect was
brought about by the most hydrated ion in the series.
Citrate gt; Tartrate gt; Sulphate gt; Acetate gt; Chloride gt;
Nitrate gt; Chlorate
while iodides and thiocyanates do not produce a turbidity in at-
tainable concentrations.

The above series holds for neutral orweakly alkaline sols. In weak-
ly acid sols however (as were first pointed out by Posternakquot;
and Pauli * in experiments on white of egg sols) the anion series
is reversed, being:

•nbsp;Lewith, Arch, experim. Pathol v. Phcrmakol 24 1 (1888); Hofmeister ibid
24 247 (1888).

In the theory we have put forward this might be explained as
being due to the hydration of the ion being imparted to the par-
ticle. Somewhat similar facts are found for the salting out and
increase of solubility of amino acids, the same ionic series varying
in the same way when neutral and acid amino acids are compa-
red

It has been stated by numerous workers that for the swelling
of gels in salt solutions the lyotropic influence makes itself felt.
The H- ion concentration, however, has a very great influence on
the degree of swelling and as no measurements are available in
which the swelUng in the different salt solutions has been com-
pared at constant H\' ion concentration, the results must be accep-
ted with caution. The lyotropic influence is most marked with anion
series. Gels seem to swell more strongly in thiocyanate « and iodide
solutions than in chloride solutions, while the swelling in chloride
solutions is more marked than in sulphates. From Hofmeister\'s
experiments 3 and also Wo. Ostwald\'s ^ experiments it appears
that sulphates hinder swelling while chlorides, nitrates and bro-
mides assist.

The lyotropic series of the cations appears to be less pronounccd.
As has already been mentioned the swelling depends greatly
on the H- ion concentration. From the earlier work it would appear
that for swelling with different acids the lyotropic series appears.
According to Loeb\' however this difference disappears when
allowance is made for the difference in the H¹ ion concentration.

Carpenter« carried out measurements of the influence of salts
on the optical rotation of gelatin. Although the experiments were

. carried out at a constant hydrogen ion concentration (Ph = 6.0)
a marked lyotropic series was found. The greatest lowering of the
optical activity was brought about by KI, the effect of KBr being
less, and that of KCl stiU less. This is in direct opposition to the
view of Loeb. According to Loeb we should expect under these
circumstances no specific anion influence, the cation alone exer-
ting an influence on the alkali side of the isoelectric point.

Gortner, Hoffmann and Sinclair ^ found a lyotropic series for
the peptisation of proteins of wheat flour by equal concentrations
of different salt solutions. Thus they found that for normal solutions
of the salts KF peptised 13 %. KCl 23 %, KBr 36 %, and KI 64 %
of the protein, the anion series being:

F lt; SO4 lt; CI lt; Tartrate lt; Br lt; I
A distinct cation series was also found although this was not so
marked.

Na lt; K lt; Li lt; Ba lt; Sr lt; Mg lt; Ca
(Here again wc meet with the Na and K ions in unexpected order)
Measurements of the hydrogen ion concentration showed that
the scries did not arise from differences in this. The results show
in a striking way how inadequate is the usual definition of globu-

^\'limilar results were obtained by von Weimam = for the pepti-
sation of cellulose, fibroin, chitin. casein, fibnn and keratin by
salt solutions for all of which he found the dispersing power of the

LiCNS gt; Lil gt; LiBr gt; LiCl
NaCNS gt; Nal
Ca(CNS), gt; Cal, gt; CaBr, gt; CaCl,

We have seen then that the same lyotropic series occur for phe-
nomena taking place in colloidal solution as for those taking place
in true solution. The spreading of these series, as shown in the
flocculation values of various sols is dependent on the hydration of
the coUoidal particles. This in the case of the vanadium pentoxide
sol the spreading is greatly diminished on the addition of alcohol
The well known example of the sulphur sol of Sven Oden (lyophilic)
where a large lyotropic influence is found and that of von Weimarn
(lyophobic) where the lyotropic influence is small may be cited
as another example ^

^^ In a case of true flocculation of a solution (as distinct from the
quot;saltmg outquot; of a sol such as agar by the addition of a lar^e quan-
tity of salt) we find that for a negative sol the least hydrated cation
has the greatest influence (where monovalent ions are considered).
We have shown that the reason for this is that hydrated oppositely
charged ions can hydrate the particle to some extent and so add to
its stability with the result that a larger quantity of salt is required
to bring about flocculation. Evidence that this view is correct is
obtained from the experiments which show the rehydrating of a
dehydrated gelatin sol by the addition of sufficient electrolyte, re-
sulting in a quot;zonequot; of flocculation, and also from the relative\'size
and appearance of the precipitates obtained by flocculating a va-
nadium pentoxide sol with lithium, sodium and potassium chlo-
rides respectively.

The case of the positive sol is different. Here the anion series
shows that the most hydrated ion has the greatest effect. Though it
naturally occurs to one that this might be due to the hydrated ions
exerting a salting out effect, we do not believe this to be the expla-
nation. If thb salting out influence of the ions made itself apparent.

there seems every reason to expect that this would be more appa-
rent in the case of the cation influence on negative sols than in the
case of the anion influence on positive sols — so that the series would
be just the reverse to what we find.

Consequently we have put forward the suggestion that the se-
quence of the anions for positive sols depends on another property
of the ion, its tendency to go into solution, which for the anion in-
creases with decreasing hydration — this, we suggest exerts a kind
of buoyancy to the colloidal particle, and so by counteracting the
hydrating effect of the ion brings about a reversal of the series.
A similar influence of the cations on negative sols would in this
case exist, but as here it would be greatest for the most hydrated
ion, it would act in the same direction as the hydration effect and
so not bring about a reversal of the series.

Thus for positive sols we have two opposing lyotropic influ-
ences. The result of adding alcohol will therefore be more diffi-
cult to foretell than in the case of the negative sol. Thus for in-
stance if the influence of the alcohol was to reduce the hydrating
effect of the ion without reducing the peptising effect (or reducing
this to a lesser degree) we would get a spreading of the lyotropic
series in the flocculation values instead of the flocculation values ap-
proaching one another. As a matter of fact we found a spreading
of the series in the case of the aluminium hydroxyde sol, and no
change in the case of the ferric hydroxide sol.

A true salting out effect (taking place in the bulk of the solution)
is to be found in the salting out of a lyophilic sol and here also the
lyotropic series appears.\' This however is quite distinct from the
lyotropic influence (which is exerted at the surface of the colloidal
particle) for the flocculation values of the less lyophilic sols.

Similar lyotropic influences of electrolytes are exerted on other
colloidal phenomena. These however need further investigation
before conclusions can be drawn. The results at present available
are however sufficient to show that a lyotropic series frequently
appears which cannot-be accounted for by differences in hydrogen
ion concentration as Loeb believed.

have been found to exist for a number of phenomena both in true
and colloidal solutions. In none of these is its occurence due to
changes in hydrogen ion concentration.

(2)nbsp;In the case of the solubility influence and alUed phenomena
the pnmary cause of the series is the salting out influence brought
about by the hydration of the ions.

(3)nbsp;In solubiUty influences we find however a specific cation
or anion influence for certain substances which cannot be ex-
plained in this way. At the extreme end of the series (correspon-
ding to low hydration of ions) even an increase in solubility some-
times occurs.

(4)nbsp;This has been shown to be due to the fact that ions orientate
water molecules and can, under certain circumstances, by bringing
about a favourable orientation actually increase the solubility of
a nonelectrolyte. This effect is superimposed on the true salting
out influence.

(5)nbsp;The presence of a hydrated ion also influences the equili-
brium nH,0 (H,0)„. From this probably arises the lyotropic
influence of the ions in displacing the maximum density of water.

(6)nbsp;The salting out of a typically lyophilic sol by a large quantity
of salt is a dehydrating effect closely connected with the solubiUty
mfluence in true solution. The lyotropic influence occurs however
not only in this phenomenon but in the flocculation values of lyo-
phobic and slightly lyophiUc colloids, where precipitation is brought
about by a comparatively small quantity of electrolyte — in this
case the lyotropic influence is more marked the more hydrated the

(7)nbsp;Dehydrating the sol with alcohol in some cases reduces the
lyotropic influence.

(8)nbsp;For negative sols the more hydrated the cation, the less its
influence in bringing about flocculation. This is because the op-
positely charged hydrated ions by introducing favourably orienta-
ted water molecules, increase the hydration of the colloidal par-
ticles.

(9)nbsp;For positive sols the more hydrated the anion the greater
its influence in bringing about flocculation. We suggest that
this is because there is a peptising influence of the anion which is
greatest for the least hydrated ion. A similar influence exists for
the cations and negative sols, but as here it increases with increasing
hydration of the ion it does not reverse the series.

(10)nbsp;A number of other colloidal phenomena where the lyotropic
series occurs are briefly considered. Differences in hydrogen ion
concentration are not the cause of the series occurring, as Loeb be-
lieved.

Krishnamurti claims that during the formation of a gel from a
gelatin sol, the colloid particles grow in size at the expense of mole-
cularly dispersed gelatin in the intermicellar liquid. He has not

The Budde effect in chlorine is not dependent on the presence
of moisture.nbsp;^^^ ^^^^^ ^^^^^ ^nbsp;,, ,48 (1929)

The e.xperiments of Stringfellow
and Duncan arc not justified in assuming that two molecular types

Sugden\'s calculations of the parachors of organic substances
have shown that Lowry\'s statement that most double bonds are
semi-polar, and that in organic chemistry a double bond usually
reacts as if it contained one co-valency and one electro-valency,
is inaccurate.

i * \' quot; \'
\'S;-, ■■ - ï^-;-\'

« *

■ i
I -, ■ •

f. •. ■
■ ■ /

; y ; . V, \' » ■

/ -v.

a\'

SOLUTION	Determinations of C. S. T.	Mean value of C.S.T.	Increase in C.S.T.
O.IM Li CI.	(77-9, 77-9)	(77-8. 77-8)	77.9°	11-4°
„ Na CI.	{73-8, 73-8)	(73-7, 73-7)	73-8	7.3
„ K CI.	(72-4, 72-6)	(72-4, 72-6)	72-5	6-0
„ Rb CI.	(72-4, 72-3)	(72-3, 72-3)	72-3	5.8
H2O	(66-5, 66-6)	(66-5, 66-5)	66-5	—
0-05M K,SO,	(75-5, 75-6)	(75-5, 75-4)	75-5	9-0
0-lM K CI.	(72-5. 72-5)	(72-5, 72-5)	72-5	6-0
„ KBr.	(72-2, 72-2)	(72-2, 72-2)	72-2	5.7
» KCIO3	(70-6, 70-6)	(70-6, 70-6)	70-5	4-1
„ KNO3	)70-6. 70-4)	(70-4, 70-5)	70-5	4-0
.. KI	(70-4, 70-6)	(70-4, 70-6)	70-5	4-0
„ KCNS	(67-7, 67-8)	(68-0. 68-0)	67-9	1-4
0-05M KCl.	(70-2, 70-2)	(70-2, 70-2)	70-2	3-7
0-1OM KCl.	(72-5, 72-5)	(72-5) 72-5,	72-5	6-0
0.15M KCl.	(75-0, 75-1)	(75-0, 75-1)	75-1	8-6
0-20M KCl.	(77-3. 77-3)	(77-0, 77-0)	77-2	10-7
0.1 MKCl.
0.00IN HCl.	(72-8. 72-8)	(72-8, 72-8)	72.8	6-3
0-05N HCl.	(69-4, 69-5)	(69-4, 69-5)	69-5	3-0
0-1 ON HCl.	{71-2, 71-3)	(71-0, 71-1)	71-2	4-7
0-15N HCl.	(73-3, 73-3)	(73-3, 73-3)	73-3	6-8
0-20N HCl.	(74-6. 74-8)	(74-6, 74-8)	74.7	8-2
O.IM HBr.	(71.1, 71-0)	(71-0. 71-0)	71-0	4-5
0.05N H2SO4	(69-3. 69-3)	(69-2. 69-2)	69-3	2-8
0.1 ON H2SO4	(71-5. 71-5)	(71-4. 71-5)	71-5	5-0
0.20N H2SO4	(74-4, 74-6)	(74-4, 74-6)	74-5	8-0
0.05M K^SO^	(75-5, 75-6)	(75-5, 75-4)	75.5	9-0
0.1 OM K2SO4	(81-4. 81-4)	(81-5. 81-3)	81-4	14-9
O.ION NaOH.	(46.5, 46.5)	(46.5, 46.5)	46.5	—20.0

50-62	74-8
54-09	80-0
47-91	70-8
	67^0
52-72	(JM = 78 or
	56-6
	59-1
	95-0

Salt Solution	Solubility in grams per litre	Solubility as a percen- tage of that in water.
Water (1-5 gs. Quinone)	13-97 ] 13-97 J	1 13-97	100-0
Water (5gs. Quinone)	13-98 \' 13-98	13-98	—
0.15 M KCNS.....	23-75 1 23-78 1	1 23-77	170-1
.. KI........	20\'86 1 20-92 i	• 20-89	149-6
quot; KNO3......	18-44 1 18-47 1	18-46	132-1
KBr.......	15-12 1 15-18 j	15-15	108-5
» KCl....... 1-5 2 M K2SO4.....	12-61 1 12-64 j 8-89 1 8-91 ƒ	12-63 8-90	90-4 63-7
= 1-5 M LiCl......			• 77-3
^ „ NaCl .....			80-3
\' » RbCl ..... „ CsCl......			93-8 (18°C) 97-6 (18X)

Substance titrated	ccs. of 20% sodium citrate added	ccs. of titanous chloride required, (about 0.05n)
25ccs. of iron alum (0\'025n)	none	2875
tt	none	28.75
f)	none	28-75
9t	30	28-93
9f	30	28-94
1»	30	28-97
it	50	29-05
	50	29\'05

Substance titrated	ccs. of sodium citrate added.	ccs. ofTiCl, required	% nitro- phenol	ccs. of Ti CI, corrected by subtracting 0.2cc.for eve- ry 30ccs. of citrate used	% nitro- phenol
25-00 CCS. of
p. nitrophenol solution	30	23-78	101-0	23-58	100-1
tgt;	30	23-75	100-9	23-55	100-0
If	60	24-01	102-0	23-61	100-2
n	60	24-01	102-0	23-61	100-2

Substances titrated	Sodium bicarbonate added	Titanous chloride required
25-00	ccs. m-nitroaniline
	solution	0-50 grams	27-64 CCS.
	ft	1-00 „	27-69	It
	n	4-00 „	27-74	II
	li	4-00 „	27-77	II
25-00	ccs. picric acid
	solution	none	21-95	II
	n	0.25 ..	21-95	II
	ft	5-00 ..	22-07	II
	gt;gt;	5-00 „	22-08	»»

Solution	ccs. of TiCl, used (corrected for citrate error)	Solubility grams per litre	Solubility as a percen- tage of that in water
Water (with 0-40gs.)	21-S7 \\ o, 83	0-5735
Water (with 0-40gs.)	21-78 1
Water (with 0-15gs.)	21-54	21-54	0-5657	100-0
	21-54
0-2M KI	23-06	23.05	0-6053	105-5
0-2M KI	23-05
0.2M KBr	22-14 \'	22-21	0-5834	101-7
0-2M KBr	22-28
0-2M KCl	21-38	21.34	0-5606	97-7
0-2M KCl	21-30
0-2M K.,SOi	20-68	20.67	0-5430	94-7
2	20-65
0-2M NaCl	20-81	20-86	0-5480	95-5
0-2M NaCl	20-91 ,
0-2M LiCl	20-72	1 20.67	0-5430	94-7
0-2^1 LiCl	20-62	1

Solution	ccs. of TiCl, used (corrected for citrate error)	Solubility In grams per litre	Solubility as a percen- tage of that in water
Water (with small excess of solid) ff	45-25 j 45-20 J	1 45-23	11-82	100
Water (with large excess of solid) tp	45-27 45-23	45-25	11-83
0-2M KI 0-2M KI	47-52 47-72	47-62	12-45	105-3
0.2M KBr 0-2M KBr	45-62 45-55	45-59	11-91	100-8
0-2M KCl 0.2M KCl	44-17 44-23	44-20	11-55	97-7
	43-96 44-04	44-00	11-50	97-3
0-2M NaCl 0-2M NaCl	42-27 42-31	42-29	11-05	93-5
0-2M LiCl 0-2M LiCl	41-69 41-73	41-71	10-90	92-2

Solution	, ccs. Ba(OH),	ccs. picirc acid combined (expressed in ccs. of Ba(OH),)	Solubilitj 1 in grams per litre	Solubility as percen- tage of that in water.
Water (with 0-25g)	22-50	43-50	! 43-50	47-23
Water (with 0-25g)	22-50	43-50	100-0
Water (with 0-5g.)	22-15	44-20	1 44-20	47-98
Water (with 0-5g.)	22-15	44-20	—
0-2M KI	*20-60	47-60	} 47-60
0-2M KI	♦20-70	47-60	51-68	109-4
0-2M KBr	21-80	44-90
0-2M KBr	21-86	44-78	i 44-84	48-68	103-0
0-2M KCl	*22-80	43-20
0-2M KCl	*22-75	43-30	} 43-25	46-85	99-2
0-2 yM K^SO,	24-40 24-30	39-70 \' 39-90	■ 39-80	43-21	91-5
0-2M NaCl	22-60	43-50
0-2M NaCl	22-40	43-50	43-50	47-23	100-0
0-2M LiCl.	♦22-15	44-50	44-50	48-32
0-2M LiCl	*22-15	44-50	102-3

Solution	Temp.	Solubility in Water (mois per litre)	Solubility in 1 mol of salt.	Sol. in 1 mol. of salt as per- centage of sol. in water
IM LiCl	17-85°	0-7219	0-6040	83-68
NaCl	17-85	0-7219	0-7090	98-23
„ KCl	17-85	0-7219	0-7694	106-6
„ RbCl	18-00	0-7319	0-7956	108-7
„ CsCl	18-00	0-7307	0-7923	108-4
„ KI	17-85	0-7219	0-7233	100-2

Solution	Solubility in	Solubility as
grams per litre	percentage
	of that in water
Water	2-12	100-0% 99-1
IM LiNOg	2-10
quot; NaNOg	1-92	90-6
quot; KNO3	1-96	92-5
quot; RbNOa	2-17	102-3
quot; CSNO3	(2-54)	119-8
quot; KI	2-23	105-2
quot; KBr	l-;79	84-5
quot; KCl	1-60	75-5
m K2SO4	1-33	62-8

É	Pt	à • Kl	or • KCNS	e	c.
• KI	• KI		•nbsp;KI •nbsp;KNOj	• Cgt;NOj	•KI
	• KDr	• KDr •UCl NaCl	• KBr	•nbsp;KI •nbsp;UbNOj	•Knr
•nbsp;KDr •nbsp;Ka	• KQ	• KCl	•CiCI •nbsp;uba •Ka •Noa •nbsp;ua	•LiNOj ,KNOj •NoNO, • KUr	• KCl »HKtSO*
•nbsp;NaQ •nbsp;ua	•NaQ •ua		•XKjSQ»	• KO
• KXiSOt ^KiS04	•HKiSO«		•NoCI

	SO,	Cl	Br	NO,
Li		0-0286	0-0271	0-0192
Na	0-0357	0-0228	0-0184	0-0l\'76
K	0-0418	0-0210	0-0151	0-0158
NH.	0-0326	0-0195	—	0-0174

tfj (in seconds)	a
0-0000	, 81-8
0-0007	80-0
0-0040	77-2
0W80	74-6
00	73-8

	M/16	M/8	M/4	M/2	Molecular Lowering
HCl......		0-7	1-3	2-6	5-2
LiCl......	—	—	1-4	2-8	5-6
NaCl.....	—	1-6	3-1	6-2	12-4
KCl......	—	1-4	2-8	5-5	11-0
NHA. ....	—	1-0	1-8	3-6	7-2
HBr.....	_	0-9	1-8	3-7	7-4
LiBr.....	—	_	1-9	3-8	7-6
NaBr.....	—	1-8	3-7	7-4	14-8
KBr.....	—	1-6	3-2	6-5	13-0
NH.Br ....	—	1-2	2-3	4-7	9.4
hi......		1-2	2-2	_	8-8
Lil......	—	1-2	2-3	—	9-2
Nal......	1-0	2-0	4-0	—	16-4
ki.....	0-9	1-8	3-7	—	14-8
nhj.....	0-7	1-4	2-7	—	10-8
HNO3.....	0-8	1-6	3-1		12-4
LiN03......	_	1-6	3-1	_	12-4
NaNOa.	1-3	2-5	5-0	_	20-0
KNO3.. ;. .	m	2-2	4-5	—	18-0
^H,N03. . . .	0-9	1-8	3-6	—	14-4
	2-0	4-0	—	—	32-0

21-87	21-92
21-88	21-92
27-57	27-65

Salt	Temperature	Viscosity	Investigator
LiCl..........	17-6°	1-147	Arrhenius
NaCl.........	17-6°	1-093
KCl..........	17;6°	0-987
RbCl..........	25°	0-9846	#1 Wagner
CsCl..........	25°	0-9775
K2SO4.........	1:7-6°	1-101	t9 Arrhenius
KNO3 .........	17-6°	0-959
KI...........	17-6°	0-912	II II

	1 vol. of sol.	1 vol. of sol.
	3 vols of water	3 vols, of alcohol.
KCNS.........	16	3-65
KI..........	54	5-0
KNO3.........	44	3-60
KCl...........	29	2-20
KCNS (repeated)......	17

	1 vol. sol. 3 vols. H,0	1 vol. sol. 1 vol. H,0 2 vols, alcohol	1 vol. sol. 3 vols, alcohol	1 vol. sol. 4 vols, alcohol
kcns . . .	\'500	500	gt; 850	gt; 680
KI.....	440	525	gt; 500
KBr ....	470	360	gt; 300
KCl ....	380	180	95	70
KCl (8 days	•		100
later) . . .	360

KCNS in mil-
limols . . .	1360 680	136	68 34 13-6	6-8 3-4
flocculation. .	— 1 —	1 -	4-	— —
appearance . .	perfectly clear, not opalescent	cloudy, no ppte.	precipitate	slightly opalescent

	69
	Table 25
(I)	Time	1 p)
Opaque throughout	0 hrs.	Opaque throughout
Top 3mm. clear. Bottom 3mm. clear.	0-75 hrs.	No change at top. Bottom 2mm. clear.
Top 6mm. clear. Bottom 4mm. clear.	3-25 hrs.	No change at top. Bottom 2mm. clear.
Top 15mm. clear. Bottom 5mm. clear.	12 hrs.	Almost clear zone at top. Bottom 8mm. clear (the whole clearing slightly).
Top 17mm. clear. Bottom 7mm. clear, (about 75 % of the salt dissolved.)	24 hrs.	Bottom 9mm. clear, above this a band of thick precipitate 7mm. wide, and abo- ve that clear liquid, (all salt dissolved).

Percentage alcohol	55%	50%	45%	40%
KCl.....	(60 2-0)	(27—2-9)	(18—4-5)	( 9-5 )
KBr.....	(70—2-6)	(30—3-3)-	(20-4-0)	(10—4-5)
KCNS.....	(80-^-0)	(35—4-0)	(22—3-5)	(10—4-0)
LiCl......	(60 2-0)	—	(18—4-7)	—

120	45 22.5 11.25 5.13	2.25 1.125 m. mois KCl.
All became as clear as water.	Opalescent and no precipitate

175 87.5 44 35	17.5 1 8.25 5.8 3.5 1.75 1.20	m. mois KCl.
All became as clear as water.	Slightly cloudy

180 90	45	36 18 1 9	5.6 3.6 1.8 1.0 1 m. mois LiCl.
All became as clear as water	Slightly cloudy